future technologies, today's choices

Future Technologies, To d a y’s ChoicesN a n o t e c h n o l o g y, Artificial Intelligence and Robotics; A technical, political and institutional map of emerging technologies.

Alexander Huw Arnall

Department of Environmental Science and Te c h n o l o g yEnvironmental Policy and Management GroupFaculty of Life SciencesImperial College LondonUniversity of London

A report for the Greenpeace Environmental Tr u s t

July 2003

Greenpeace Environmental Tr u s t

Canonbury VillasLondon N1 2PN

www.greenpeace.org.uk

ISBN 1-903907-05-5

Published by Greenpeace Environmental Tr u s t

Canonbury Villas, London N1 2PN

ISBN 1-903907-05-5

Printed on 100% recycled paper

1Future Technologies, Today's Choices

List of Ta b l e s 2

Abbreviations and Acronyms 3

F o r e w o r d 4Dr Doug Parr, Greenpeace Chief Scientist

A c k n o w l e d g e m e n t s 9

1 . I n t r o d u c t i o n1 . 1 About nanotechnology,

a rtificial intelligence and robotics 101 . 2 R e p o rt structure 101 . 3 Key references 11

2 . N a n o t e c h n o l o g y2 . 1 I n t r o d u c t i o n 122 . 1 . 1 About nanotechnology 132 . 1 . 2 Where are we now? 13

2 . 2 Research and Development2 . 2 . 1 I n t r o d u c t i o n 142 . 2 . 2 Novel materials 142 . 2 . 3 N a n o t u b e s 142 . 2 . 4 Tools and fabrication 162 . 2 . 5 Public funding for

research and development 18

2 . 3 Applications and Markets2 . 3 . 1 I n t r o d u c t i o n 212 . 3 . 2 I n f o r m a t i c s 222 . 3 . 3 Pharmaceuticals and medicine 252 . 3 . 4 E n e r g y 272 . 3 . 5 D e f e n c e 292 . 3 . 6 Corporate funding 31

2 . 4 Reality and Hype2 . 4 . 1 I n t r o d u c t i o n 322 . 4 . 2 Molecular nanotechnology 332 . 4 . 3 Fundamental barriers to these visions 35

2 . 5 C o n c e r n s2 . 5 . 1 I n t r o d u c t i o n 352 . 5 . 2 Environmental concerns 362 . 5 . 3 Socio-political concerns 372 . 5 . 4 Public acceptance of nanotechnology 392 . 5 . 5 The regulation debate 40

2 . 6 D i s c u s s i o n 41

3. Artificial Intelligenceand Robotics

3 . 1 I n t r o d u c t i o n 423 . 1 . 1 About AI and robotics 423 . 1 . 2 Where are we now? 42

3 . 2 Aspects of Research3 . 2 . 1 I n t r o d u c t i o n 433 . 2 . 2 L e a r n i n g 443 . 2 . 3 Reasoning about plans,

programs and action 453 . 2 . 4 Logical AI 453 . 2 . 5 C o l l a b o r a t i o n 453 . 2 . 6 P e r c e p t i o n 463 . 2 . 7 Human–computer interaction 463 . 2 . 8 Public funding 46

3 . 3 A p p l i c a t i o n s3 . 3 . 1 I n t r o d u c t i o n 493 . 3 . 2 Intelligent simulation systems 493 . 3 . 3 Intelligent information resources 503 . 3 . 4 Intelligent project coaches 503 . 3 . 5 R o b o t i c s 523 . 3 . 6 Corporate funding 53

3 . 4 Reality and Hype3 . 4 . 1 I n t r o d u c t i o n 543 . 4 . 2 Barriers to strong AI 543 . 4 . 3 A future for strong AI? 55

3 . 5 C o n c e r n s3 . 5 . 1 I n t r o d u c t i o n 563 . 5 . 2 Predictive intelligence 563 . 5 . 3 AI and robotic autonomy 57

3 . 6 D i s c u s s i o n 59

4 . C o n c l u s i o n 60

E n d n o t e s 62

R e f e r e n c e s 63

C o n t e n t s

2

Table 1: Summary of the major nanomaterials currently in research anddevelopment and their potential applications.

Table 2: Applications for new materials and devices resulting from self-assemblyand self-organisation.

Table 3: World-wide government funding for nanotechnology research and development.

Table 4: Breakdown of spending on the US’s National Nanotechnology Initiative from 2001–2003.

Table 5: Top five government spending on nanotechnology in the Far East in 2002.

Table 6: Estimated Japanese government nanotechnology research and development expenditures.

Table 7: Top six European government nanotechnology spending from 1998–2000.

Table 8: Summary of future estimated global markets in nanotechnology.

Table 9: Anticipated technological computing developments for 2001–2014.

Table 10: Maturity of lithography options.

Table 11: Summary of application areas for informatics.

Table 12: Summary of application areas for nanoscale pharmaceuticals and medicine.

Table 13: Summary of applications for energy processing.

Table 14: US historical funding for technology transitioning into the marketplace.

List of Ta b l e s


AAAI American Association of Artificial Intelligence

AI artificial intelligence

ANN artificial neural network

ASIMO Advanced Step in Innovative Mobility

CBEN Centre for Biological andEnvironmental Nanotechnology

CMOS complementary metal oxidesemiconductor

CNID Centre for Nanoscience Innovation for Defence

DARPA Defence Advanced Research Project Agency

DoD Department of Defence

DRAM dynamic random access memory

DTI Department of Trade and Industry

DDT dichlorodiphenyltrichloroethane

EC European Commission

EU European Union

EELD Evidence Extraction and Link Discovery

EPA Environmental Protection Agency

EPSRC Engineering and Physical Sciences Research Council

FP Framework Programme

GM genetically modified

ISS Intelligent Simulation System

IT information technology

MEMS micro-electrical-mechanical systems

METI Ministry of Economy,Trade and Industry

MEXT Ministry of Education,Culture, Sports, Science and Technology

MIT Massachusetts Instituteof Technology

MNT molecular nanotechnology

NASA National Aeronautics andSpace Administration

NBIC nanoscience, biotechnology,information technology andcognitive science

NII National Institute of Informatics

NNI National Nanotechnology Initiative

NSF National Science Foundation

PC personal computers

PV photovoltaic

QIP quantum information processing

RAM random access memory

RWCP Real World Computing Project

SCI Scientific Citation Index

TIA Total Information Awareness

UCAV Unmanned Combat Air Vehicle

A b b r e v i a t i o n sand Acronyms

4

Why is Greenpeace interested in newtechnologies? New technologies featureprominently in our ongoing campaignsagainst genetic modified (GM) crops andnuclear power; however, they are also anintegral part of our solutions toenvironmental problems, including renewableenergy technologies, such as solar, wind andwave power, and waste treatmenttechnologies, such as mechanical–biologicaltreatment. So while Greenpeace accepts andrelies upon the merits of many newtechnologies, we campaign against othertechnologies that have a potentially profoundnegative impact on the environment.

Greenpeace is in the business of evaluatingboth future and current threats. Our missionmust be to survey upcoming innovations forseveral reasons. First, we are conscious ofunintended (but foreseeable) consequencesthat impact on the environment. No oneintended, for example, that pesticide use inthe 1970s and 1980s would have the impacton wildlife that it did. Becoming aware of,and ultimately preventing, the environmentaldownside of technological developments isclearly a core interest – indeed, the‘precautionary principle’ has become animportant part of international law, such asthe Biosafety Protocol on GM organisms.There is also increasing interest in the widerconcept of precaution, which is nowrecognised to include the need for widerparticipation in the control and direction oftechnological innovation. This kind ofprocess produces not only a better evidencebase, but also more informed decisions.Unintended consequences of a particular newtechnology cannot always be foreseen;however, if these consequences become acollective problem, it is unreasonable toexpect collective responsibility if the decisionto proceed with the technology was made byan elite few.

Second, and more subtly, the interests of thosewho own and control the new technologies

l a rgely determine how a new technology isused. Any technology placed in the hands ofthose who care little about the possiblee n v i ronmental, health, or social impacts ispotentially disastrous. When entire nationaleconomies are adapted to take advantage ofthe economic opportunities off e red by newtechnologies, it is a matter of huge publici m p o rtance, and the potential enviro n m e n t a land social consequences are clearly ofi m p o rtance to Greenpeace. Globaltechnologies can, particularly in the long term ,be of greater significance than Prime Ministersor presidents. Will the power aff o rded topeople and organisations in control of thesenew technologies be properly controlled? If asingle person – a computer- v i rus writer or abiochemist dealing with anthrax – can causehuge political and financial problems, howmuch more damage could those with morere s o u rces do? Thorough public scrutiny beforefinancial or political commitments to newtechnologies become irreversible could behugely beneficial, and surely a matter ofdemocratic rights.

In April and May 2002, Greenpeace andNew Scientist magazine co-sponsored a seriesof four debates on the impacts of newtechnologies, entitled Science, Technology

and the Future. These debates generatedmuch interest, but the difficulties in locatingspeakers highlighted the fact that few peoplecould give an overview of eitherdevelopments in these technologies or theirimpact in the physical, political andcommercial domains. Even more problematicwas identifying what the initial technologicalproducts would be and their social orenvironmental consequences.

This prompted Greenpeace to commission acomprehensive review of nanotechnology andartificial intelligence/robotics developmentsfrom an organisation with a reputation fortechnological expertise – Imperial CollegeLondon. We asked them to documentexisting applications and to analyse current

F o r e w o r dDr Doug Parr, Greenpeace Chief Scientist


research and development (R&D), the mainplayers behind these developments, and theassociated incentives and risks.

New Technologies in Context

Beyond the contents of this report, thepolitical and social processes surrounding theintroduction of technologies are veryimportant. For example, compare the publicresponse to GM crops in Europe to the wideacceptance of mobile phones. The ‘socialconstitution’ of the technology appears keyto its acceptability. This social constitutionprovides the answers to questions such as:

• Who is in control?

• Where can I get information that I trust?

• On what terms is the technology beingintroduced?

• What risks apply, with what certainty, andto whom?

• Where do the benefits fall?

• Do the risks and benefits fall to the samepeople (e.g. mobile phones are popular,while mobile phone masts are not)?

• Who takes responsibility for resultingproblems?

The evidence presented in this report suggeststhat, depending on the development pathway,some aspects of nanotechnology might get arocky ride, as its social constitution is morelike that of GM crops than mobile phones. Inparticular, future disputes surrounding newtechnology seem certain in the light ofglobalised, rapid technology transfer. Thegeneral public is also increasingly unwillingto accept the word of a company orGovernment (on the basis of brutalexperience), on the risks and benefits oftechnology, particularly as science andcommerce become more closely linked.

At the time of commissioning this report,civil society critiques of the immense R&Dand commercial efforts taking place innanotechnology were quite sparse, butalready there are signs that this is changing.In the wake of the furore over geneticmodification, the idea of a ‘public debate’about new technologies is in vogue, but thishas to be meaningful or it will simplypromote cynicism.

If public dialogue on science is to meananything, the approach of nanotechnology isa huge opportunity. Instead of waiting forpotential adverse reactions, the scientificcommunity could be proactive. Why not holda citizens jury to determine scientificpriorities on nanotechology? From each ofthe agricultural, defence, energy,pharmaceutical, and information technology(IT) sectors (and the numerous cross-overs),the jury could examine current research andits potential. It could suggest which areasneed to be highest priority. It would look atthe potential short- and long-termapplications and the ‘blue skies’ elementnecessary for any research programme.Research councils such as the Biotechnologyand Biological Sciences Research Council(BBSRC) and the Engineering and PhysicalSciences Research Council (EPSRC) in theUK could commit to considering results andutilizing the insights from the findings ofsuch a jury. If dialogue between science andsociety is to be more than just a sophisticatedmeans of engineering user-acceptance,research councils must adopt this kind ofparticipatory initiative to allow ordinarypeople to have a say in the types andtrajectories of technological innovation.

N a n o t e c h n o l o g y

The most common definition ofnanotechnology is that of manipulation,observation and measurement at a scale ofless than 100 nanometres (one nanometre isone millionth of a millimetre). However, theemergence of a multi-disciplinary field called

6

‘nanotechnology’ arises from newinstrumentation only recently available, anda flow of public money into a great numberof techniques and relevant academicdisciplines in what has been described as an‘arms race’ between governments.Nanotechnology is really a convenient labelfor a variety of scientific disciplines whichserves as a way of getting money fromGovernment budgets. The figures involvedare becoming very large; indeed this reportindicates that over US$2 billion was spent bynational governments in 2002, and that thesefigures will be even larger in 2003. Althoughthe US is said to be the leader, the Japanesegovernment is expected to spend more thanthe US in 2003. It is also thought that 2002will prove to be the year when corporatefunding matched or exceeded state funds.This is because transnational companiesrealise that nanotechnology is likely todisrupt their current products and processes,and because the investment community hasdecided that nanotechnology is the ‘next bigthing’. Three new business alliances haverecently been formed in the US, Europe andAsia, whose sole purpose is to translateresearch into economically viable products.The UK Government’s Department of Tradeand Industry estimates that the market fornanotechnology applications will reach overUS$100 billion by 2005. There is now agreat deal of momentum behindnanotechnology that has built up into a forcewhich might already struggle to incorporatethe outcomes of organised public debate, ormeet well-founded public concerns, althoughby no means will all of the developments becontroversial – many will not.

The difficulty in making predictions aboutthe future is that R&D could still takeseveral different directions, and the materialsand processes being developed aretechnology-pushed rather than market-led.After the hype about possible applications,the first real nanotechnology products arestarting to appear in the semiconductor

industry – to increase storage densities onmicrochips – and in the pharmaceuticalindustry to improve drug targeting anddiagnostic aids. Both sectors expect that inthe future nanotechnology will provide adramatic leap forward, but that for now theproducts seem relatively modest compared tothe preceding hype. Other areas of futureapplications appear to be within the energysector and defence. With regard to theformer, more effective solar cells and highlyefficient lighting hold promise on a ten-yeartime-scale. In the latter, there is no shortageof ideas for military applications and at leasttwo new institutions in the US have beencreated expressly for the purpose ofexploiting nanotechnology for military gain.

Notice that none of these applications dealwith the far more distant but highly-publicised prospect of replicator robots orthe so-called ‘general assembler’ – a nano-machine which would produce anythingdesired given the right raw-materials, andwhich formed some of the ideas behindMichael Crichton's novel, Prey. Theseapplications are currently a long way off dueto the difficulties involved in engineeringchemical building blocks, informationmanagement, and systems design. Thechallenges are formidable but even so, twoUS companies are known to be researchingmolecular assembly. The ‘runaway replicator’concerns (also known as the ‘grey goo’scenario) raised by Crichton’s novel arehideous, but the prospects of it remain wayoff, and some experts suggest that it wouldbe very difficult to achieve this deliberately,let alone by accident (but see below).

All of this suggests that the development ofnanotechnology will go through variousdifferent stages, and thus societal debate willneed to be an ongoing process rather than asingle outcome. There will need to becontinual incorporation of the insights fromsuch a debate into policy and productdevelopment as the prospects become more


tangible. Already some concerns arebecoming evident. Some new materials mayconstitute new classes of non-biodegradablepollutant about which we have littleunderstanding. Additionally, little work hasbeen done to ascertain the possible effects ofnanomaterials on living systems, or thepossibility that nanoparticles could slip pastthe human immune system. Carbonnanotubes are already found in cars andsome tennis rackets, but there is virtually noenvironmental or toxicological data on them.Despite this, of the US$710 million beingspent by the US Government onnanotechnology, only US$500,000 is beingspent on environmental impact assessment,even though a major feature of the productpipeline is that it consists of new materials.Current proposals at EU level on syntheticchemicals regulation are belatedly ensuringthat a rule of ‘no data, no market’ will applyto the basic information about hazardousproperties of such chemicals. Knowing thebasics about the dangers of new materials isa pre-requisite for effective environmentalresponsibility. From the Greenpeaceperspective, this suggests that whilst ‘societaldebate’ is highly desirable, it is a bit of aluxury if the same old mistakes are beingrepeated by a new generation oftechnologists. There is no need for grand,new mechanisms of public involvement topoint out the blindingly obvious. With causefor concern, and with the precautionaryprinciple applied, these materials should beconsidered hazardous until shown otherwise.

Still other concerns are evident in the socialarena that revolve around the uses to whichthe new technology is put – closely linkedwith ownership and control. One possibledystopian future would be the shift of thecontrol of nanotechnology towards beingdriven by military needs. This report doesnot generally support such a prospect atpresent, although military interest innanotechnology is considerable.Alternatively, corporate control has been

flagged up by the ETC group, and thisimplies the pursuit of income streams fromthose already possessing disposable income.Is the future of nanotechnology then, aplaything of the already-rich? Will the muchtalked about ‘digital-divide’ be built upon,exacerbating the inequities present in currentsociety through a ‘nano-divide’?Nanotechnology can only be made availableto the poor and to developing countries if thetechnology remains open to use. Already acompany in Toronto has applied for patentson the carbon moleculeBuckminsterfullerene. If ownership ofmolecules is allowed, the nanotechnologytechniques for the precise manipulation ofatoms open up a whole new terrain forprivate ownership. As with geneticengineering where genes have becomecontrolled by patents, things that were onceconsidered universally owned could becomecontrolled by a few.

A rtificial Intelligence and Robotics

Unlike the situation for nanotechnology,researchers in artificial intelligence (AI) feelthat their work has suffered because of‘public discussion’ – hype might be a betterterm – in the 1960s and 1980s whichadversely affected advances in the field afterthe delivery did not live up to expectationsand funding dropped. Many researchers nowfeel that the goal of mimicking the humanability to solve problems and achieve goals inthe real world (the so-called ‘strong AI’) isneither likely nor desirable because a longseries of conceptual breakthroughs isrequired. Instead the focus is on ‘weak AI’ –applications that model some, but not all,aspects of human behaviour.

The number of applications for weak AI isgrowing. AI-related patents in the USincreased from 100–1700 between 1989 and1999, with a total of 3900 patentsmentioning related terms. AI systems aregenerally embedded within larger systems –applications can be found in video games,

8

speech recognition, and in the ‘data mining’business sector. Full speech recognition,leading to voice-led Internet access orrecognition in security applications, isanticipated relatively soon. However, theability to extract meaning from naturallanguage recognition remains way off. Thedata mining market uses software to extractgeneral regularities from online data, dealingin particular with large volumes or patternshumans may not look for. Such systemscould be used to predict consumerpreferences or extract trends from marketdata such as patents and news articles. Salesalready have reached US$3.5 billion and areanticipated to be US$8.8 billion in 2004.Weak AI is already behind systems thatdetect ‘deviant’ behaviour in credit card use,which has lead to improved credit card frauddetection. Potential applications of thesetechniques to state-security situations arelikely to be controversial (see below).

The field of robotics is closely linked to thatof AI, although definitional issues abound.‘Giving AI motor capability’ seems areasonable definition, but most people wouldnot regard a cruise missile as a robot eventhough the navigation and control techniquesdraw heavily on robotics research. After thehype from the 1960s rebounded oninvestment (as for AI), experts moved awayfrom the idea of complete automation as itwas neither desirable nor feasible. Instead,more practical applications have been found,such as in cervical smear screening and,predictably, in the military sphere, whereUnmanned Combat Air Vehicles (UCAVs) arebeing developed, with the hope of fieldingthem by 2008.

Despite these developments, current AIsystems are, it is argued, fundamentallyincapable of exhibiting intelligence as weunderstand it. Current AI is only as smart asthe programmer who wrote the code. AIs o f t w a re designers point out that existingcomputer arc h i t e c t u re means that most AI

applications necessarily arise through classicaldesign and programming techniques, ratherthan new approaches that aim to allowp rogrammes to train and evolve. An exampleof such an alternative approach may bepossible through artificial neural networks,although these systems are so complex that itis not generally possible to follow thereasoning processes that they exhibit.

The funding of AI research is far moredifficult to uncover than for nanotechnologyas no existing overview seems to exist on thetopic, and information on spending is usuallyplaced under a general computer sciencebudget. Industry reportedly leads, with two-thirds of spending on research in computerscience, even though public spending hasproved an important source of funding in thepast, largely because of the field’s high-riskconceptual challenges. Nevertheless it is clearthat the US is the leader in spending. It leads,in part, due to military-related institutions,such as the Defence Advanced ResearchProject Agency (DARPA) and the NationalAeronautics and Space Administration(NASA) who used AI systems and roboticsfor the exploration of Mars. Japan andEurope are also investing (and indeedcollaborating) in this field, but are playingcatch-up with the US, although Japanremains the leader in using industrial robots.

Far more likely than the tyrannical take-overof society by hyper-intelligent robots (afrequent science fiction theme) or concernsabout ‘rights’ for intelligent machines, amore likely issue will be the use of AIsystems to spy on people. The USDepartment of Defense has established agroup to look at information gathering andanalysis on a huge scale, includinggovernment and commercial sources, whichwould use AI systems to scrutinise the dataand extract information about people,relationships, organisations, and activities forcounter-terrorism purposes. The concernsabout infringing personal privacy or possible


misuse of the data are clear. Furthermore, theuse of computer systems for the US NationalMissile Defence, and possibly for UCAVs,has created a different moral dilemma in that“they will be the first machines given theresponsibility for killing human beingswithout human direction or supervision”.

AI and robotics are likely to continue to cre e pinto our lives without us really noticing.U n f o rt u n a t e l y, many of the applicationsappear to be taking place amongst agencies,p a rticularly the military, that do not re a d i l yrespond to public concern, however wella rticulated or thought through.

The Future

Nanotechnology and AI/robotics, togetherwith biotechnology, may well be on aconvergent path. In 2001 the NationalScience Foundation held a large workshop tolook at the implications of this convergenceand the implications for human abilities andproductivity. AI could be boosted bynanotechnology innovations in computingpower. Applications of a futurenanotechnology general assembler wouldrequire some AI and robotics innovations.

Equally, nanotechnology may converge muchsooner with biotechnology as it uses the toolsand structures of biological systems togenerate tiny machines. Although theprospect of general assemblers may be quitedistant, self-replicating ‘machines’ that usethe tools of biology – and look more likeliving things than machines – might be closerat hand through the convergence of bio- andnanotechnologies. ‘Grey goo’ might not be arealistic prospect; ‘green goo’ may be closerto the mark – quite how close is difficult tojudge on the basis of the evidence in thisreport. Any creation that posed the prospectof being self-replicating would need to behandled with immense care to ensureenvironmental protection.

Whether any of the technological futuresbeing scoped out in laboratories are whatour general public would like is a questionthat can only be answered by asking them. Ifthose concerned with the development ofnew technologies, and nanotechnology inparticular, are convinced that the benefitsthey hope to generate will withstand scrutinythey should have no concerns about engagingand winning public support.

Many thanks to my supervisors, TimothyFoxon and Robert Gross, Imperial CollegeLondon, for their guidance and advice incompleting this report; to Douglas Parr,Greenpeace, for commissioning the work;and to Ken Green, University of ManchesterInstitute of Science and Technology, for hisreview and commentary.

In addition, I would to thank Gareth Parry,Jenny Nelson, and Murray Shanahan ofImperial College London; Abid Khan of theLondon Centre for Nanotechnology; and

Olivier Bosch of the International Institutefor Strategic Studies (IISS) for allowing me tointerview them.

Finally, I am grateful to Jon Glick of theAmerican Association for ArtificialIntelligence (AAAI); Andre Gsponer of theIndependent Scientific Research Institute(ISRI); Hope Shand of the ETC Group; andLoretta Anania, Ramon Compano, andJakub Wejcher of the European Community’sFuture and Emerging Technologiesprogramme (EC FET) for their assistance.

A c k n o w l e d g e m e n t s

10

1.1 About nanotechnology,artificial intelligence and robotics The aim of this report is to provide basic,background information of global scope onthree emerging technologies: nanotechnology,artificial intelligence (AI) and robotics.According to the Department of Trade andIndustry (DTI), it is important to considerthese emerging technologies now becausetheir emergence on the market is anticipatedto ‘affect almost every aspect of our lives’

during the coming decades (DTI, 2002).Thus, a first major feature of these threedisciplines is product diversity. In addition, itis possible to characterise them as disruptive,

enabling and interdisciplinary.

D i s ruptive technologies are those that displaceolder technologies and enable radically newgenerations of existing products and pro c e s s e sto take over. They can also enable whole newclasses of products not previously feasible.The implications for industry are considerable:companies that do not adapt rapidly faceobsolescence and decline, whereas those thatdo sit up and take notice will be able to donew things in almost every conceivabletechnological discipline (DTI, 2002).Nanotechnology is also an enablingtechnology and, like electricity, the intern a lcombustion engine, or the Internet, its impacton society will be broad and oftenunanticipated. Unlike these examples,h o w e v e r, nanotechnology is generallyc o n s i d e red harder to ‘pin down’ – it is ageneral capability that impacts on manyscientific disciplines (Holister, 2002). Inaddition, the interd i s c i p l i n a ry features of thesenew technologies result in another drivingfactor for innovation and discovery: they canbring together people from traditionallyseparate academic groups. For example, theboundaries between physical sciences and lifesciences are blurring within these fields.

1.2 Report structureThis report is divided in two main parts: thefirst examines the field of nanotechnology,and the second looks at AI and robotics.Furthermore, both parts are divided into sixequivalent sections. The Section 1 of eachpresents an introduction. Following this, thecurrent status of research and development(R&D) is described for both fields in Section2, with particular attention being paid to theareas of research attracting the mostattention. Much of the work described herecuts across traditional academic boundariesand contains a significant technical element.This is because a firm understanding of thenature of the technology itself is essential inunderstanding its future impact (Holister,2002). In addition, the perspective taken hereis global in scope since governments andcorporations world-wide are investing inthese areas and research is active on severalcontinents. This suggests that, withinternational flows of information,technological innovation will betransboundary in nature.

The applications and markets of theseemerging technologies are described inSection 3. Specifically, this report aims tohighlight the kinds of products which havealready been introduced into the globalmarket and those applications due forintroduction in the short- and medium-term.In addition, the range of market values thatare currently being anticipated are pointedout, although these figures are necessarilyhighly speculative. Underpinning these R&Dand application developments is a wide arrayof key players. While interest in thesetechnologies is increasing rapidly, particularlyin nanotechnology, most of the recent growthof interest comes from those with a strategicinterest, such as governments, venturecapitalists, large technology-orientatedcorporations and scientists working in thefield (Holister, 2002).

1. Introduction


One problem with many of the hundreds ofdocuments written about emergingtechnologies every year is that they do notdistinguish between science and sciencefiction, let alone the desirable andundesirable in terms of ethics, choice andsafety (Ho, 2002b). Thus, Sections 4 and 5aim to deal with some of these issues: Section4 separates out some of the hype from themore visionary but solidly placedapplications, whereas Section 5 provides anaccount of the potential environmental andsocial risks that such uses could pose in thefuture. Finally, Section 6 highlights some ofthe key messages of each part.

1.3 Key references This report has been compiled by consultinga wide variety of sources across the entirespectrum of the debate, from industryadvocates to environmental and socialpressure groups. In doing so, a number ofsources have been particularly important. Forthe section on nanotechnology, the DTI’s(2002) New Dimensions for Manufacturing:

UK Strategy for Nanotechnology provides auseful introduction to the field. In addition,Ramon Compano (2001) of the EuropeanCommission; Professors J.N. Hay and S.J.Shaw (2000) of the University of Surrey andDefence Evaluation and Research Agency(DERA); Paul Holister (2002) of CMPCientifica; Ian Miles and Duncan Jarvis(2001) of the National Physical Laboratory(NPL); and Ottilia Saxl (2000) of theInstitute of Nanotechnology have been usedextensively for construction of summarytables. Finally, the National ScienceFoundation (NSF) report, Societal

Implications of Nanoscience and

Nanotechnology, supplies good informationon a wide range of issues (Roco andBainbridge, 2001). For the section on AI andRobotics, Barbara Grosz and Randall Davis– President and President-Elect of theAmerican Association for ArtificialIntelligence (AAAI) – and Daniel Weld of theUniversity of Washington provide someuseful technical information.

12

2.1 Introduction

2 . 1 . 1 About nanotechnology

A major difficulty of characterisingnanotechnology is that the field does notstem from one established academicdiscipline (The Economist, 2002). In fact,there are a number of ways in whichnanotechnology may be defined. The mostcommon version regards nanoscience as ‘the

ability to do things – measure, see, predict

and make – on the scale of atoms and

molecules and exploit the novel properties

found at that scale’ (DTI, 2002).Traditionally, this scale is defined as beingbetween 0.1 and 100 nanometres (nm), 1 nmbeing one-thousandth of a micron(micrometre; mm), which is, in turn, one-thousandth of a millimetre (mm). However,as will become clear in the later stages of thisstudy, this definition is open tointerpretation, and may readily be applied toa number of different technologies that haveno obvious common relationship (TheEconomist, 2002).

Another way to characterise nanotechnologyis by distinguishing between the fabricationp rocesses of top-down and bottom-up. To p -down technology refers to the ‘fabrication of

nanoscale stru c t u res by machining and

etching techniques’ (Saxl, 2000). However,top-down means more than justminiaturisation: at the nanoscale leveld i ff e rent laws of physics come into play,p ro p e rties of traditional materials change,and the behaviours of surfaces start todominate the behaviour of bulk materials.On the other hand, bottom-up technology –often re f e rred to as molecularnanotechnology (MNT) – applies to thec reation of organic and inorganic stru c t u re s ,atom by atom, or molecule by molecule(Saxl, 2000). It is this area of nanotechnologythat has created the most excitement andp u b l i c i t y. In a mature nanotech world,m a c ro s t ru c t u res would simply be grown fro mtheir smallest constituent components: an

‘anything box’ would take a molecular seedcontaining instructions for building a pro d u c tand use tiny nanobots or molecular machinesto build it atom by atom (Miller, 2002).Indeed, as Forrest (1989) points out, ‘ t h e

development of [bottom-up] technology does

not depend upon on discovering new

scientific principles. The advances re q u i re d

a re engineering.’ In short, fully-fledgedbottom-up nanotechnology promises nothingless than complete control over the physicals t ru c t u re of matter – the same kind of contro lover the molecular and structural makeup ofphysical objects that a word pro c e s s o rp rovides over the form and content of text(Reynolds, 2002).

2 . 1 . 2 Where are we now?

At present it is clear that this bottom-up‘dream’ is far from being realised. As Saxl(2000) notes: ‘Top-down and bottom-up can

be a measure of the level of advancement of

nanotechnology, and nanotechnology, as

applied today, is still mainly in the top-down

stage.’ This state of relative infancy is oftencompared in the literature to the informationtechnology (IT) sector in the 1960s, orbiotechnology in the 1980s. So, with thescience fiction aspects of the debate rapidlyreceding, industry has now necessarilyadopted much more realistic expectations(pers. comm., Abid Khan, London Centre forNanotechnology, 6 Nov 2002.)

This is not to say, however, that we havelong to wait before nanotechnology makes itsmark in the global market. In fact, currentindustry jargon would probably describenanotechnology as ‘coming on stream’. For,although the underlying technologies andtheir applications are still at an early stage ofdevelopment, there are applications emerginginto the market that are likely to be makinga significant impact on the industrial sceneby 2006 (Miles and Jarvis, 2001). The bestevidence of this move into commercialisationconcerns the recent emergence of threealliances whose sole purpose is to translate

2. Nanotechnology


this underlying research into economicallyviable products: the US NanoBusinessAlliance, the Europe NanobusinessAssociation, and the Asia-PacificNanotechnology Forum. In addition to this,laboratories around the world are workingon new approaches and on new ways to scaleup nanotechnology to industrial levels. Forexample, the first factories to manufacturecarbon nanotubes and fullerenes are underconstruction in Japan (DTI, 2002).

In spite of these developments, there hasbeen criticism recently over the amount of hype and, consequently, funding thatresearch into nanoscience andnanotechnology has received. For example,the much-heralded US NationalNanotechnology Initiative (NNI) has beencriticised for using ‘nano’ as a convenient tagto attract funding for a whole range of newscience and technologies (e.g. see Roy, 2002).This reinvention is one way of attractingmore money because politicians like to feelthey are putting money into something newand exciting (pers. comm., Gareth Parry,Imperial College London, 22 Nov 2002). For these reasons, the nanotechnology sectoris far broader than you would usually expectto see and the resulting lack of a cleardefinition is hampering meaningfuldiscussion of its potential costs or benefits.Thus, if we use the standard definition givenabove, we can say that nanoscience andtechnology have been around for severaldecades, particularly in research,development, and manufacturing in IT.Rather, it is the wide availability of tools andinformation to diverse scientific communitiesthat has generated the current interest in thisarea (Chaudhari, 2001).

2.2 Research and Development

2 . 2 . 1 I n t r o d u c t i o n

The absence of a universally accepted strictdefinition of nanotechnology has allowed theresearch emphasis to broaden, encompassingmany areas of work that have traditionallybeen referred to as chemistry or biology(DTI, 2002). Thus, the first majorcharacteristic of activity grouped under thissection is that contemporary R&D cutsacross a wide range of industrial sectors. In some cases, major markets are fairly welldefined. The food industry serves as a goodexample here, where there are significantdrivers at work (pers. comm., Abid Khan,London Centre for Nanotechnology, 6 Nov2002). To illustrate, ‘smart’ wrappings forthe food industry (that indicate freshness orotherwise) are close to the market (Saxl,2000). By 2006, beer packaging isanticipated by industry to use the highestweight of nano-strengthened material, at 3 million lbs., followed by meats andcarbonated soft drinks. By 2011, meanwhile,the total figure might reach almost 100million lbs. (nanotechweb.org, 2002). Inother cases, important applications areidentified but the eventual market impactsare more difficult to predict. For example,nanotechnology is anticipated to yieldsignificant advances in catalyst technology.If these potential applications are realisedthen the impact on society will be dramaticas catalysts, arguably the most importanttechnology in our modern society, enable theproduction of a wide range of materials andfuels (Saxl, 2000).

A second characteristic of current work inthis area is that the kinds of materials andprocesses being developed are necessarily‘technology pushed’: urged on by thepotential impacts of nanotechnology, theR&D community is achieving rapid advancesin basic science and technology. This level ofscientific interest is gauged by Compano andHullman (2001) who examine the world-

14

wide number of publications innanotechnology in the Science Citation Index(SCI) database. They conclude that for theperiod between 1989 and 1998 the averageannual growth rate in the number ofpublications is an ‘impressive’ 27%. This risein interest is not confined to a small numberof central repositories however (Smith,1996). Instead, research is spread acrossmore than 30 countries that have developednanotechnology activities and plans (Holister,2002). In this way, Compano and Hullman(2001) also examine the distribution of thisinterest. Based upon their findings, the mostactive is the US, with roughly one-quarter ofall publications, followed by Japan, China,France, the UK and Russia. These countriesalone account for 70% of the world’sscientific papers on nanotechnology. Inparticular, for China and Russia the sharesare outstanding in comparison with theirgeneral presence in the SCI database andshow the significance of nanoscience in theirresearch systems.

2.2.2 Novel materials

The third major characteristic of activitygrouped under this section concerns that factthat nanotechnology is primarily aboutmaking things (Holister, 2002). For thisreason, most of the existing focus of R&Dcentres on ‘nanomaterials’: novel materialswhose molecular structure has beenengineered at the nanometre scale (DTI,2002). Indeed, Saxl (2000) states that:‘material science and technology is

fundamental to a majority of the applications

of nanotechnology.’ Thus, many of thematerials that follow (Table 1) involve eitherbulk production of conventional compoundsthat are much smaller (and hence exhibitdifferent properties) or new nanomaterials,such as fullerenes and nanotubes (ETCGroup, 2002a). The markets range ofnanomaterials are considerable. Indeed, it has been estimated that, aided bynanotechnology, novel materials andprocesses can be expected to have a market

impact of over US$340 billion within adecade (Holister, 2002).

2.2.3 Nanotubes

Nanotubes provide a good example of howbasic R&D can take off into full-scalemarket application in one specific area.Described as ‘the most important material in

nanotechnology today’ (Holister, 2002),nanotubes are a new material withremarkable tensile strength. Indeed, takingcurrent technical barriers into account,nanotube-based material is anticipated tobecome 50–100 times stronger than steel atone-sixth of the weight (Anton et al., 2001).This development would dwarf theimprovements that carbon fibres brought tocomposites. Harry Kroto, who was awardedthe Nobel Prize for the discovery of C60Buckminsterfullerene, states that suchadvances will take ‘a long, long time’ toachieve (2010 Nanospace Odyssey lecture,Queen Mary University, 6 Jan 2003), the firstapplications of nanotubes being in compositedevelopment. However, if such technologiesdo eventually arrive, the results will beawesome: they will ‘be equivalent to James

Watt’s invention of the condenser’, adevelopment that kick-started the industrialrevolution. The concept of the space elevatorserves as a good illustration of the kind ofvisionary thinking that recent nanotubedevelopment has inspired. The idea of a ‘liftto the stars’ is not itself particularly new: aRussian engineer, Yuri Artutanov, penned theidea of an elevator – perhaps powered by alaser that could quietly transport payloadsand people to a space platform – as early as1960 (cited in Cowen 2002). However, suchideas have always been hampered by the lackof material strength necessary to make thecable attachment. The nanotube may be thekey to overcoming this longstandingobstacle, making the space elevator a realityin just 15 years time (Cowen, 2002). Thisdevelopment, though, will rely on thesuccessful incorporation of nanotubes intofibres or ribbons and successfully avoiding

Table 1: Summary of the major nanomaterials currently in research and development and their potential applications.

M a t e r i a l P r o p e rt i e s A p p l i c a t i o n s Time-scale (to

market launch)

Clusters of atoms

Quantum wells Ultra-thin layers – usually a few nanometres thick – CD players have made use of quantum Current – 5 yearsof semiconductor material (the well) grown between well lasers for several years. More barrier material by modern crystal growth technologies recent developments promise to make (Saxl, 2000). The barrier materials trap electrons in the these nanodevices commonplace in ultra-thin layers, thus producing a number of useful low-cost telecommunications and optics.properties. These properties have led, for example, to the development of highly efficient laser devices.

Quantum dots Fluorescent nanoparticles that are invisible until ‘lit up’ Telecommunications, optics. 7–8 yearsby ultraviolet light. They can be made to exhibit a range of colours, depending on their composition (Miles and Jarvis, 2001).

P o l y m e r s Organic-based materials that emit light when an electric Computing, energy conversion. ?current is applied to them and vice versa (pers. comm., Jenny Nelson, Imperial College London, 2 Dec 2002).

Grains that are less than 100nm in size

N a n o c a p s u l e s Buckminsterfullerenes are the most well known Many applications envisaged Current – 2 yearsexample. Discovered in 1985, these C60 particles are e.g. nanoparticulate dry lubricant1nm in width. for engineering (Saxl, 2000).

Catalytic nanoparticles In the range of 1–10 nm, such materials were Wide range of applications, including Current – ?in existence long before it was realised that they materials, fuel and food production,belonged to the realms of nanotechnology. health and agriculture (Hay and H o w e v e r, recent developments are enabling a given S h a w, 2000).mass of catalyst to present more surface area forreaction, hence improving its performance (Hay and S h a w, 2000). Following this, such catalytic nanoparticlescan often be regenerated for further use.

Fibres that are less than 100nm in diameter

Carbon nanotubes Two types of nanotube exist: the single-wall carbon Many applications are envisaged: space Current – 5 yearsnanotubes, the so-called ‘Buckytubes’, and multilayer and aircraft manufacture, automobilescarbon nanotubes (Hay and Shaw, 2000). Both consist and construction. Multi-layeredof graphitic carbon and typically have an internal carbon nanotubes are already availablediameter of 5 nm and an external diameter of 10 nm. in practical commercial quantities. Described as the ‘most important material in Buckytubes some way off large-scalenanotechnology today’ (Holister, 2002), it has been commercial production (Saxl, 2000).calculated that nanotube-based material has the potentialto become 50–100 times stronger than steel at one sixthof the weight.

Films that are less than 100nm in thickness

S e l f - a s s e m b l i n g Organic or inorganic substances spontaneously form A wide range of applications, based 2–5 yearsmonolayers (SAMs) a layer one molecule thick on a surface. Additional on properties ranging from being

layers can be added, leading to laminates where each chemically active to being wear layer is just one molecule in depth (Holister, 2002). resistant (Saxl, 2000).

N a n o p a r t i c u l a t e Coating technology is now being strongly influenced Sensors, reaction beds, liquid crystal 5–15 yearsc o a t i n g s by nanotechnology. E.g. metallic stainless steel manufacturing, molecular wires,

coatings sprayed using nanocrystalline powders lubrication and protective layers, anti-have been shown to possess increased hardness corrosion coatings, tougher and harder when compared with conventional coatings (Saxl, 2000). cutting tools (Holister, 2002).

Nanostructured materials

N a n o c o m p o s i t e s Composites are combinations of metals, ceramics, A number of applications, particularly Current – 2 yearspolymers and biological materials that allow multi- where purity and electrical conductivityfunctional behaviour (Anton et al., 2001). When characteristics are important, such as materials are introduced that exist at the nanolevel, in microelectronics. Commercial nanocomposites are formed (Hay and Shaw, 2000), exploitation of these materials is and the material’s properties – e.g. hardness, currently small, the most ubiquitoust r a n s p a r e n c y, porosity – are altered. of these being carbon black, which finds

widespread industrial application, particularly in vehicle tyres(Hay and Shaw, 2000).

Te x t i l e s Incorporation of nanoparticles and capsules into M i l i t a r y, lifestyle. 3-5 yearsclothing leading to increased lightness and durability, and ‘smart’ fabrics (that change their physicalproperties according to the wearer’s clothing) ( H o l s t e r, 2002).

16

various atmospheric dangers, such aslightning strikes, micrometeors, and human-made space debris.

The market impetus behind suchdevelopments, then, is clear: the conventionalspace industry is anticipated as the firstmajor customer, followed by aircraftmanufacturers. However, as production costsdrop (currently US$20–1200/g), nanotubesare expected to find widespread applicationin such large industries as automobiles andconstruction. In fact, it is possible toconceive of a market in any area of industrythat will benefit from lighter and strongermaterials (Holister, 2002). It is expectationssuch as these that are currently fuelling therace to develop techniques of nanotube mass-production in economic quantities. The ETCGroup (2002b) states that there are currentlyat least 55 companies involved in nanotubefabrication and that production levels willsoon be reaching 1 kg/day in somecompanies. For example, Japan’s Mitsui andCo. will start building a facility in April 2003with an annual production capacity of 120 tons of carbon nanotubes (Fried, 2002).The company plans to market the product toautomakers, resin makers and batterymakers. In fact, the industry has grown soquickly that Holister (2002) believes that thenumber of nanotube suppliers already inexistence are not likely to be supported byavailable applications in the years to come.Fried (2002) also supports this contention,stating that the ‘carbon nanotube field is

already over-saturated’.

2 . 2 . 4 Tools and fabrication

It is a simple statement of fact that in order tomake things you must first have thefabrication tools available. There f o re, many ofthe nanomaterials covered above are co-evolving with a number of enablingtechnologies and techniques. These toolsp rovide the instrumentation needed toexamine and characterize devices and eff e c t sduring the R&D phase, the manufacturing

techniques that will allow the larg e - s c a l eeconomic production of nanotechnologyp roducts, and the necessary support forquality control (DTI, 2002). Because of theessential nature of this category, its influence isfar greater than is reflected in the size of theeconomic sectors producing these pro d u c t s .For this reason, the tools and techniqueshighlighted below have a strong commerc i a lf u t u re and the greatest number of establishedcompanies (pers. comm., Gareth Parry,Imperial College London, 22 Nov 2002). Thefollowing sections cover methods for top-down and bottom-up manufacture, softwaremodelling and nanometro l o g y. However, inthe near future, this area will mainly featureextensions of conventional instru m e n t a t i o nand top-down manufacturing. More futuristicmolecular scale assembly remains distant(Miles and Jarvis, 2001).

2.2.4.1 Top-down manufacture

Scanning Probe Microscope. This is thegeneral term for a range of instruments withspecific functions. Fundamentally, ananoscopic probe is maintained at a constantheight over the bed of atoms. This probe canbe positioned so close to individual atomsthat the electrons of the probe-tip and atombegin to interact. These interactions can bestrong enough to ‘lift’ the atom and move itto another place (pers. comm., Gareth Parry,Imperial College London, 22 Nov 2002).

Optical Te ch n i q u e s . These techniques can beused to detect movement – obviouslyi m p o rtant in hi-tech precision engineering.Optical techniques are, in theory, restricted inresolution to half the wavelength of lightbeing used, which keeps them out of the lowernanoscale, but various approaches are nowo v e rcoming this limitation (Holister, 2002).

Lithographics. Lithography is the means bywhich patterns are delineated on silicon chipsand micro-electrical-mechanical systems(MEMS). Most significantly, opticallithography is the dominant exposure tool in


use today in the semiconductor industry’sComplementary Metal Oxide Semiconductor(CMOS) process

2.2.4.2 Bottom-up manufacture

The tools here support rather more futuristicapproaches to large-scale production andnanofabrication based on bottom-upapproaches, such as nanomachine productionlines (Miles and Jarvis, 2001). This approachis equivalent to building a car engine up fromindividual components, rather than the lessintuitive method of machining a systemdown from large blocks of material. Indeed,although such techniques are still in theirinfancy, the DTI (2002) report a recentmovement away from top-down techniquestowards self-assembly within theinternational research community. Scientistsand engineers are becoming increasingly ableto understand, intervene and rearrange theatomic and molecular structure of matter,and control its form in order to achievespecific aims (Saxl, 2000).

Self-assembly and self-organisation. Self-assembly refers to the tendency of somematerials to organise themselves into orderedarrays (Anton et al., 2001). This techniquepotentially offers huge economies, and isconsidered to have great potential innanoelectronics. In particular, the study ofthe self-assembly nature of molecules isproving to be the foundation of rapid growthin applications in science and technology. Forexample, Saxl (2000) reports that theStranski–Krastonov methods for growingself-assembly quantum dots has rendered thelithographic approach to semiconductorquantum dot fabrication virtually obsolete.In addition, self-assembly is leading to thefabrication of new materials and devices. Theformer area of materials consists of newtypes of nanocomposites or organic/inorganichybrid structures that are created bydepositing or attaching organic molecules toultra-small particles or ultra-thin manmadelayered structures (Hay and Shaw, 2000).

Similarly, the latter area of devices rangefrom the production of new chemical and gassensors, optical sensors, solar panels andother energy conversion devices, to bio-implants and in vivo monitoring. The basisof these technologies is an organic film (theresponsive layer) which can be deposited ona hard, active electronic chip substrate. Thesolid-state chip receives signals from theorganic over-layer as it reacts to changes inits environment, and processes them. Theapplications for these new materials anddevices are summarised in Table 2.

2.2.4.3 Software Modelling

Molecular modelling software is anotherfabrication technique of wide-rangingapplicability as it permits the efficientanalysis of large molecular structures andsubstrates (Miles and Jarvis, 2001). Hence, it is much used by molecular

Table 2: Applications for new materials and devices

resulting from self-assembly and self-organisation.

N a m e Te c h n i q u e A p p l i c a t i o n

New materials

Sol-gel technology Inorganic and The design of different(Miles and Jarvis, 2001) organic component types of materials;

c o m b i n a t i o n . functional coatings.

Intercalation of Intercalation of Toxicity testing, drugpolymers (Miles and polymers with other delivery and drug Jarvis, 2001) materials (DNA, drugs). performance analysis.

N a n o - e m u l s i o n s Nanoparticle size and Production of required(Saxl, 2000) composition selected. viscosity and absorption

c h a r a c t e r i s t i c s .

B i o m i m e t i c s Design of systems, High strength, structural(Anton et al., 2001) materials and their applications, such as

functionality to mimic artificial bones andnature. t e e t h .

New devices

Field-sensing devices Combination of Biosensing and(Saxl, 2000) molecular films with optical switching.

optical waveguidesand resonators.

M a t e r i a l - s e n s i n g Surfaces of liquid Gas and chemical devices (Saxl, 2000) crystals or thin sensing.

membranes and otherorganic compoundscan be used to detectmolecules via structuralor conductive changes.

18

nanotechnologists, where computers cansimulate the behaviour of matter at theatomic and molecular level. In addition,computer modelling is anticipated to proveessential in understanding and predicting thebehaviour of nanoscale structures becausethey operate at what is sometimes referred to as the mesoscale, an area where bothclassical and quantum physics influencebehaviour (Holister, 2002).

2.2.4.4 Nanometrology

Fundamental to commercial nanotechnologyis repeatability, and fundamental torepeatability is measurement.Nanometrology, then, allows the perfectionof the texture at the nanometre and sub-nanometre level to be examined andcontrolled. This is essential if highlyspecialised applications of nanotechnologyare to operate correctly, for example X-rayoptical components and mirrors used in lasertechnologies (Saxl, 2000).

2 . 2 . 5 Public funding for

research and development

The main reason for government interest in nanotechnology is strategic: to achieve an advantageous position so that whennanotech applications begin to have asignificant effect in the world economy,countries are able to exploit these newopportunities to the full. Harper (2002), whodescribes the current situation as a global‘arms race’, puts these ideas into perspective:

‘You only have to look at how IT made a

huge difference to both the US economy and

US military strength to see how crucial

technology is. Nanotechnology is an even

more fundamental technology than IT. Not

only has it the ability to shift the balance of

military power but also affect the global

balance of power in the energy markets.’

This emphasis on military power is wellfounded: Smith (1996) echoes this sentimentwhen he speculates that it is entirely possible

that much, or even most, US governmentresearch in the field is concentrated in thehands of military planners.

Levels of public investment innanotechnology are reminiscent of a growingstrategic interest: this is an area that attractsboth large and small countries. Global R&Dspending is currently around US$4 billion(ETC Group, 2002a), with public investmentincreasing rapidly (503% between 1997 and2002 across the ‘lead’ countries2). Table 3summarises these rises.

2.2.5.1 The US

The US is widely considered to be the world-leader in nanoscale science research (Saxl,2000). Certainly, in terms of leading centresfor nanotechnology research, the USAdominates, with eight institutions making theDTI (2002) top list of 13. These centres areUniversity of Santa Barbara, CornellUniversity, University of California at LosAngeles, Stanford University, IBM ResearchLaboratories, Northwestern University,Harvard University and the MassachusettsInstitute of Technology (MIT). In total, morethan 30 universities have announced plansfor nanotech research centres since 1997(Leo, 2001). Further, the US is widely

Table 3: World-wide government funding for nanotechnology

research and development (US$million).

A r e a 1 9 9 7 1 9 9 8 1 9 9 9 2 0 0 0 2 0 0 1 2 0 0 2 2 0 0 3

U S * 1 1 6 1 9 0 2 5 5 2 7 0 4 2 2 6 0 4 7 1 0

We s t e r nE u r o p e 1 2 6 1 5 1 1 7 9 2 0 0 2 2 5 ~ 4 0 0 N A

J a p a n 1 2 0 1 3 5 1 5 7 2 4 5 4 6 5 ~ 6 5 0 N A

O t h e r s * * 7 0 8 3 9 6 1 1 0 3 8 0 ~ 5 2 0 N A

To t a l 4 3 2 5 5 9 6 8 7 8 2 5 1 5 0 2 2 1 7 4 N A

(% of 1997) 1 0 0 1 2 9 1 5 9 1 9 1 3 4 8 5 0 3 N A

NA: not available.* Excluding non-federal spending e.g. California. ** ‘Others’ includes Australia, Canada, China, Eastern Europe, theFormer Soviet Union, Singapore, Taiwan and other countries withnanotechnology R&D. For example, in Mexico there are 20 researchgroups working independently on nanotechnology. Korea, already aworld player in electronics, has an ambitious 10-year programme toattain a world-class position in nanotechnology (DTI, 2002).


regarded as the benchmark against whichnanotechnology funding should be compared(Roman, 2002). Indeed, Howard (2002)states that, ‘while other governments are

investing in a range of nanotechnology

research, the US effort is by far the most

substantial.’ From 1985–1997 the totalsupport for projects related tonanotechnology was estimated at US$452million, coming in roughly equal parts fromthe NSF, various industrial sponsorship, andother government funding. Then in 2000, themuch-heralded NNI was launched – a multi-agency programme designed to provide a bigfunding boost for nanotechnology. There arecurrently 10 US government partners in theNNI3. These are shown in Table 4.

Table 4 shows that the NSF and Departmentof Defence (DoD) are the two majorrecipients of investment in nanoscience andtechnology R&D. Indeed, the NSF hasdesignated ‘nanoscale science and

engineering’ as one of its six priority areas,while the DoD has dedicated its funding to

elaborating a ‘conceptual template for

achieving new levels of war-fighting

effectiveness’ (DoD, 2002). This tableprovides a fairly accurate picture of currentresearch priorities in the US. However, statefunding, which can sometimes be substantial,is not included in the estimates. For example,the state of California, which is home tovirtually all the work in molecularnanotechnology, has invested US$100 millionin the creation of a California NanosystemsInstitute. And neither are the figures static;levels of funding are anticipated to increaserapidly once the economic benefits of USfunding begin to be felt, whether in newcompany start-up activity, or progresstowards military or social goals.

2.2.5.2 Far East

Table 5 shows the levels of 2002 governmentspending on nanotechnology within fivecountries in the Far East. On average, thesefigures are lower than in the US although,given the increased purchasing power incountries such as China, they may beconsidered as relatively high (Roman, 2002).However, while the figures given are up-to-date, the time-scales over which they operateare ambiguous.

Of all the countries shown in Table 5,Japan’s nanotech investments are by far thegreatest. Indeed, it is universally agreed thatJapan has the only fully co-ordinated andfunded national policy of nanotechnologyresearch. The most prominent product of this

Table 4: Breakdown of spending on the US’s National

Nanotechnology Initiative from 2001–2003 (US$million).

Recipient 2001 2 0 0 2 2 0 0 3

a c t u a l e s t i m a t e p r o p o s e d

National ScienceF o u n d a t i o n 1 4 5 1 9 9 2 2 1

Department of Defence 1 2 5 1 8 0 2 0 1

Department of Energy 7 8 9 1 1 3 9

National Aeronautics 0 4 6 4 9and Space Administration

National Institute 4 0 4 1 4 3of Health

National Institute of 2 8 3 7 4 4Standards and Te c h n o l o g y

Environmental 5 5 5Protection Agency

Department of 0 2 2Tr a n s p o r t a t i o n

US Department 0 2 5of Agriculture

Department of Justice 1 1 1

To t a l 4 2 2 6 0 4 7 1 0

DTI, 2002.

Table 5: Top five government spending

on nanotechnology in the Far East in 2002

( U S $ m i l l i o n ) .

C o u n t r y S p e n d i n g

J a p a n 7 5 0

C h i n a 2 0 0

K o r e a 1 5 0

Taiwan 1 1 1

S i n g a p o r e 4 0

To t a l 1 2 5 1

Roman, 2002.

20

national policy has been the Ministry ofEconomy, Trade and Industry (METI)programme on atomic manipulation,1991–2001, entitled Research and

Development of Ultimate Manipulation of

Molecules (Tam, 2001). The programme wasfunded at the ¥25 billion level(approximately US$210 million). Of thetotal, US$167 million has been allocated forthe development of microbots (Saxl, 2000).Nowadays, the Japanese government viewsthe successful development ofnanotechnology as key to restoration of itseconomy: nanotechnology is one of the fourstrategic platforms of Japan’s second basicplan for science and technology. Forexample, the Japanese government hasfounded the Expert Group onNanotechnology under the Japan Federationof Economic Organisations Committee onIndustrial Technology. In another initiative,which it calls its ‘e-Japan strategy’, theJapanese government aims to become ‘the

world’s most advanced IT nation within five

years’ (IT Strategic Headquarters, 2001).Japan’s government nanotechnologyexpenditures are given in Table 6.

Although the figures given in Table 6 areimpressive, Roman (2002) believes that theannual 50% increase does cast some doubtover their accuracy. For while there is nodoubt that funding will continue to increase,increasing the number of researchersavailable to absorb this extra funding doesnot seem possible on an annual basis.

2.2.5.3 European Union

All European Union (EU) member states,except Luxembourg where no universities arelocated, have re s e a rch programmes. For somecountries, such as Germ a n y, Ireland orSweden, where nanotechnology is considere dof strategic importance, nanotechnologyp rogrammes have been established for several

years. On the other hand, many countrieshave no specifically focused nanotechnologyinitiatives, but this re s e a rch is covered withinm o re general R&D programmes (Compano,2001). Table 7 summarises the situation forthe top six countries.

The European Commission (EC) fundsnanoscience through its so-called FrameworkP rogramme (FP). The aim of the FP6 is top roduce bre a k t h rough technologies thatd i rectly benefit the EU, either economically ors o c i a l l y. Under this, €1.3 billion aree a rmarked for ‘nanotechnologies and

nanosciences, knowledge-based

multifunctional materials and new pro d u c t i o n

p rocesses and devices’ in the 2002–2006 FPout of a total budget of €11.3 billion. Thisthematic priority is only partly dedicated tonanoscience, while other thematic prioritiesalso have a nanotechnology component. Atfirst glance this may seem a small figurec o m p a red to the 2003 NNI budget of US$710million (€0.72 billion). However, it does nottake into account the substantial contributionsmade by individual member states (Compano,2001). The UK serves as a good example ofthis, where public spending onnanotechnology R&D was around £30 millionin 2001 (DTI, 2002), 70–80% of it from theEngineering and Physical Sciences Researc hCouncil (EPSRC). However, this is set to risequite rapidly in 2002–2003 as the newi n t e rd i s c i p l i n a ry re s e a rch collaborations anduniversity technology centres start to spread.

Table 6: Estimated Japanese government nanotechnology research

and development expenditures (US$million).

1 9 9 7 1 9 9 8 1 9 9 9 2 0 0 0 2 0 0 1 2 0 0 2 2 0 0 3

1 2 0 1 3 5 1 5 7 2 4 5 4 6 5 ~ 7 5 0 ~ 1 0 0 0

Roman, 2002.

Table 7: Top six European government

nanotechnology spending from 1998–2000 (€m i l l i o n ) .

C o u n t r y / i n s t i t u t i o n 1 9 9 8 1 9 9 9 2 0 0 0

G e r m a n y 4 9 . 0 5 8 . 0 6 3 . 0

U K 3 2 . 0 3 5 . 0 3 9 . 0

European Commission 2 6 . 0 2 7 . 0 2 9 . 0

France 1 2 . 0 1 8 . 0 1 9 . 0

N e t h e r l a n d s 4 . 7 6 . 2 6 . 9

Sweden 3 . 4 5 . 6 5 . 8

European total 1 3 9 . 8 1 6 4 . 7 1 8 4 . 0

Compano, 2001.


2.3 Applications and Markets

2 . 3 . 1 I n t r o d u c t i o n

The applications of nanotechnology areextremely diverse, mainly because the field isinterdisciplinary (Miles and Jarvis, 2001). Inaddition, the effect that nanotechnology willhave during the next decade is difficult toestimate because of potentially new andunanticipated applications. For example, ifsimply reducing the microstructure inexisting materials can make a big marketimpact, then this may, in turn, lead to awhole new set of applications. However, itseems reasonable to assume that during thenext two to three years most activity innanotechnology will still be in the area ofresearch, rather than completed projects orproducts. Holister (2002) estimates that thereare currently 455 public and privatecompanies, 95 investors, and 271 academicinstitutions and government entities that areinvolved in the near-term applications ofnanotechnology world-wide. The ability ofsuch institutions to transfer research resultsinto industrial applications can be indicatedby the number of filed patents. Companoand Hullman (2001) provide an analysis ofthis, using the number of nanopatents filed atthe European Patent Office (EPO) inMunich. Over the whole 1981–1998 period,the number of nanopatents rises from28–180 patents, with an average growth ratein the 1990s amounting to 7%.

One important characteristic of activitygrouped within this section is that much ofthe work in near-term applications ofnanotechnologies is ‘market-pulled’: in eachcase, a particular and potentially profitableuse within industry and/or the consumermarket has been identified. However, as withthe difficulty in predicting the futureapplications of nanotechnology, manymarket analysts believe that it is too soon toproduce reliable figures for the global market– it is simply too early to say where andwhen markets and applications will come

(DTI, 2002). In spite of these difficulties,some forecasts exist that do hint at the kindof growth we might expect.

Most strikingly, the NSF predicts that thetotal market for nanotech products andservices will reach US$1 trillion by 2015(Roco and Bainbridge, 2001). The accuracyof this claim is difficult to assess, given thedoubts expressed above. Compano andHullman (2001) approach the problemthrough the comparison of publication(representing basic science or R&D) andpatent (representing technology applications)nanotechnology data with Grupp’s (1993)theory of Stylised TechnologicalDevelopment. As a result, they conclude thatthe peak of scientific activity is still to come,possibly in three to five years from now, andlarge-scale exploitation of nanotechnologicalresults might arise ten years from now.

Considering the above comments aboutnanotechnological development and market-pull, it is instructive to examine which areasof industry will be affected first. MihailRoco, the NSF senior advisor fornanotechnology, believes that ‘early payoffs

will come in computing and pharmaceuticals’

(quoted in Leo, 2001), whereas Holister(2002) points out that medicine is a hugemarket, thereby implying that revenue fornanotechnology in this area could besubstantial. On the other hand, the NSFbelieve that, due to the high initial costsinvolved, ‘nanotechnology-based goods and

services will probably be introduced earlier in

those markets where performance

Table 8: Summary of future estimated

global markets in nanotechnology.

Ye a r Estimated global market

2 0 0 1 £31–55 billion

2 0 0 5 £105 billion

2 0 0 8 £500 billion

2 0 1 0 £700 billion

2 0 1 1 – 2 0 1 5 Exceeds US$1 trillion (£0.6 trillion)

DTI, 2002.

22

characteristics are especially important and

price is a secondary consideration’ (Roco andBainbridge, 2001). Examples of these aremedical applications and space exploration.The experience gained will then reducetechnical and production uncertainties andprepare these technologies for deploymentinto the market place.

A good indication of the areas of current andnear-future commercial nanotech activity isthe type of patents made. Compano andHullman (2001) state that one-quarter of allpatents filed are focused on instrumentation.This supports the view that nanotechnologyis at the beginning of the development phaseof an enabling technology where the firstfocus is to develop suitable tools andfabrication techniques. The most importantindustrial sectors are informatics(information science), and pharmaceuticalsand chemicals. For the first sector, ‘massive

storage devices, flat panel displays, or

electronic paper are prominent IT patenting

areas. In addition to this, extended

semiconductor approaches and alternative

nanoscale information processing,

transmission or storage devices are

dominant.’ In the case of chemistry andpharmaceuticals, a large number of patentsare directed towards ‘finding new approaches

for drug delivery, medical diagnosis, and

cancer treatments which are supposed to

have huge future markets. Nanotechnology

patenting for other sectors (e.g. aerospace,

construction industries and food processing)

show yearly increasing values, but their

absolute numbers are relatively small.’ Insummary then, IT and medicine look set tohave an impact on the market first. The nexttwo sections deal with both these areas inmore detail. Following this, the widely citedpotential impacts of nanotechnology on theenergy and defence sectors are examined.

2.3.2 Informatics

Informatics, or information science, can bethought of as consisting of three interrelated

areas: electronics, magnetics and optics. Thissection primarily concentrates on electronics,acknowledged by Compano (2001) as one ofthe major drivers of the world-wideeconomy. In fact, the current market forminiaturised systems is estimated at US$40billion and the market for IT peripherals tobe more than US$20 billion, althoughsemiconductor products have a dominantrole and their turnover grows at a higher ratethan the overall electronics market. The fieldis dominated by the US and Japan. In fact,apart from a few niche markets whereWestern European companies are able tocompete, recent technological breakthroughshave been largely due to majormanufacturers in these countries (Miles andJarvis, 2001). Japan has a particularly strongcommercial basis in this area, althoughJapanese R&D tends to be organised throughlines determined by the government (via theMicroMachine Centre): the METI fundsmuch of the work (US$100 million in the lastfive years). In the US too, government is veryinvolved in applied research. Here, theactivities of military funding agencies are ofnote – such institutions tend to be generousin their company funding in this field, evenwhen there is a clear commercial benefit forthe companies involved.

In general, it is much harder to predict thecommercially successful technologies in theworld of electronics than in the world ofmaterials (Holister, 2002). However, if oneconsiders that the major driving force innanoscience for the last decade has beenmicroelectronics (Glinos, 1999), then itmakes sense that nanotechnology will playan important role in the future of thisindustry. The ETC Group (2002a) provide anotable statistic here, stating that by 2012the entire market will be dependent onnanotech. For, although there are fewnanotechnology products in the market placeat present, future growth is expected to bestrong, with a predicted composite annualgrowth rate of 30–40%, with emerging


markets around 70% (DTI, 2002). A numberof recent forecasts, although varying greatly,reflect this market confidence. For example,Miles and Jarvis (2001) put the market fornanotechnology-based IT and electronicsdevices at around US$70 billion by 2010. Asecond estimate states that nanotechnologywill yield an annual production of aboutUS$300 billion for the semiconductorindustry and about the same amount againfor global integrated circuits sales within10–15 years (NSF, 2001). Similarly, formicro- and nanotechnology systems in thetelecommunications sector, the market ispresently estimated as being in the order ofUS$35 billion with an anticipated compoundannual growth rate of around 70%.

2.3.2.1 Moore’s Law

The microelectronics industry had lookedahead and seen serious challenges for itsbasic CMOS process. CMOS technology hasbeen refined for over 20 years, driving the‘line width’-the width of the smallest featurein an integrated circuit (IC)-from 10 mmdown to 0.25 µm (Doering, 2001). This is

the force behind Moore’s law, which predictsthat the processing power of ICs will doubleevery 18 months (Glinos, 1999). Based onMoore’s law, industry predictions aresummarised in Table 9.

Semiconductor industry associations assumethat they will be close to introducing 100 nmground-rule technology by 2004 (Compano,2001). The significance of this lies in the factthat 100 nm is widely viewed as a kind of‘turning point’, where many radically newtechnologies will have to be developed. Tobegin with, optical lithography will becomeobsolete somewhere around 100 nm. As aresult, ‘next generation lithography’ optionsare currently being investigated. These aresummarised in Table 10.

Excluding the printing process, eachfabrication technique essentially works onthe same principle where a reactive silicon-based agent is exposed to increasinglyfocused electromagnetic radiation: optical toX-rays representing a successive reduction inphoton wavelength; E-beam and ion beamprojection technologies using focusedelectron and ion beams respectively. All ofthese techniques are currently under activeevaluation-the aim is to have the appropriateequipment for the corresponding time-frame.To date, X-ray and ion bean projection have

Table 9: Anticipated technological

computing developments for 2001–2014.

F e a t u r e Ye a r

2 0 0 1 2 0 0 3 2 0 0 5 2 0 0 8 2 0 1 1 2 0 1 4

M e m o r y

Minimum feature 1 5 0 1 2 0 1 0 0 7 0 5 0 3 5size DRAM (1/2 pitch in nm)

G b i t s / c h i p 2 4 8 2 4 6 8 1 9 4

Density 0 . 4 9 0 . 8 9 1 . 6 3 4 . 0 3 9 . 9 4 2 4 . 5 0( G b i t s / c m2)

Logic (processing power)

Minimum feature 1 0 0 8 0 6 5 4 5 3 0 – 3 2 2 0 – 2 2size (gate length in nm)

Density (million 1 3 2 4 4 4 1 0 9 2 6 9 6 6 4transistors per cm2)

Logic clock (GHz) 1 . 7 2 . 5 3 . 5 6 . 0 1 0 . 0 1 3 . 5

DRAM: Dynamic Random Access Memory, a type of memory used in most personal computers.

Adapted from Compano, 2001

Table 10: Maturity of lithography options.

Year of introduction 2 0 0 1 2 0 0 3 2 0 0 6 2 0 0 9

Minimum feature size 1 5 0 1 2 0 9 0 6 5

Optical 193 nm X * X

Optical 157 nm X X

Extreme UV X X

X - r a y s X

Electron beam X X

Ion beam X X

P r i n t i n g X

*An ‘X’ designates the date at which the respective fabrication technology is expected to become economically viable for mass production.

Adapted from Compano, 2001.

24

received the greatest research investment(Compano, 2001). Printing technologies,however, are the ultimate goal, where sheetsof circuits can be rolled off the productionline like a printing press.

2.3.2.2 Beyond Moore’s law

M o o re ’s law cannot continue indefinitely. In the years following 2015, additionald i fficulties are likely to be encountere d ,some of which may pose serious challengesto traditional semiconductor manufacturingtechniques. In part i c u l a r, limits to the degre ethat interconnections or wires betweentransistors may be scaled could in turn limitthe effective computation speed of devicesbecause of the pro p e rties and compatibilityof particular materials, despite incre m e n t a lp resent-day advances in these areas (Antonet al., 2001). Thermal dissipation in chipswith extremely high device-densities will alsopose a serious challenge. This issue is not somuch a fundamental limitation as it is aneconomic consideration, in that heatdissipation mechanisms and coolingtechnology may be re q u i red that add to thetotal system cost, thereby adversely aff e c t i n gthe marginal cost per computationalfunction for these devices. Eventually,h o w e v e r, CMOS technology may hit a morec rucial barr i e r, the quantum world, wherethe laws of physics operate in a veryd i ff e rent paradigm to that experienced ine v e ryday life. For example, futuristic circ u i t soperating on a quantum scale would have totake Heisenberg ’s Uncertainty Principle intoaccount. Overcoming this barrier is ad i ff e rent matter altogether, where thep roblems are no longer merely technological(Glinos, 1999), and industry has alre a d ybegun to investigate the problem in anumber of ways. Three of the mostcommonly cited appro a c h e s - m o l e c u l a rn a n o e l e c t ronics and quantum inform a t i o np rocessing (QIP)-are expanded upon below.In addition, computational self-assembly isacknowledged as a potentially keyfabrication technique of the future .

Molecular nanoelectronics. Organicmolecules have been shown to have thenecessary properties to be used in electronics.Devices made of molecular componentswould be much smaller than those made byexisting silicon technologies and ultimatelyoffer the smallest electronics theoreticallypossible without moving into the realm ofsubatomic particles (Holister, 2002).Molecular electronic devices could operate as logic switches through chemical means,using synthesised organic compounds. Thesedevices can be assembled chemically in largenumbers and organised to form a computer.The main advantage of this approach issignificantly lower power consumption byindividual devices. Several approaches forsuch devices have been devised, andexperiments have shown evidence ofswitching behaviour for individual devices.

For example, in ‘DNA computing’, thesimilarities between mathematical operationsand biological reactions are used to performcalculations. The key idea is to find theparallelism between DNA-the basic geneticinformation-and well-known digitalcomputers. This is because a string of DNAcan be used to solve combination problems if it can be put together in the right sequence(Compano, 2001). One issue is thatmolecular memories must be able tomaintain their state, just as in a digitalelectronic computer. Also, given that themanufacturing and assembly process forthese devices will lead to device defects, a defect-tolerant computer architectureneeds to be developed. Fabricating reliableinterconnections between devices usingcarbon nanotubes (or some other technology)is an additional challenge. A significantamount of work is ongoing in each of theseareas. Even though experimental progress todate in this area has been substantial, itseems unlikely that molecular computerscould be developed within the next 15 yearsthat would be relatively attractive (from aprice and performance standpoint) compared


with conventional electronic computers(Anton et al., 2001).

QIP. This crosses the disciplines of quantumphysics, computer science, informationtheory and engineering with the aim ofharnessing the fundamental laws of quantumphysics to ‘dramatically improve the

acquisition, transmission and processing of

information’ (Miles and Jarvis, 2001). QIPrepresents computing at the smallest possiblescale, in which one atom is equivalent to onebyte of information. Other aspects ofquantum computing also consideredattractive relate to their massive parallelismin computation (i.e. the ability to performsimultaneous calculations) (Holister, 2002).These concepts are qualitatively differentfrom those employed in traditionalcomputers and will hence require newcomputer architectures. A preliminary surveyof work in this area by Anton et al., (2001)indicates that quantum switches are unlikelyto overcome major technical obstacles, suchas ‘error correction, de-coherence and signal

input/output’ within the next 15 years. If thisproves to be the case, QIP-based computing,as with molecular nanoelectronics, does notappear to be competitive with traditionaldigital electronic computers for some time.

Computational self-assembly. A major barrierto the introduction of nanoelectronics is thatthere are no established mass productiontechniques for creating devices on acommercial basis (Saxl, 2000). In the shortto medium term, Table 10 covers the mostpromising fabrication approaches. In thelong-term, however, more ambitious bottom-up methods based on self-assemblytechniques are proposed. Bottom-upapproaches are elegant, cheap and possiblyenormously powerful techniques for futuremass replication. The relativelystraightforward architecture of molecularmemory means that self-assembly techniquesin this area may bear fruit in a few years(Table 11). Tackling processors is another

matter, however, because of the greatercomplexity involved-their applicabilityremains limited until total control over theemerging structures in terms of wiring andtheir interconnections can be obtained. Theseare formidable obstacles. Self-assembly,therefore, will likely be combined initiallywith some more traditional top-downapproaches. For example, many believe thatinducing molecular components to self-assemble on a patterned substrate in somesort of hybrid system will represent the firstcommercialisation of nanoelectronics(Holister, 2002).

2.3.2.3 Summary of applications

The main drivers for current appliedmicroelectronics research are computing,telecommunications, consumer electronicsand military applications. It is not evidenthow long personal computing will act as adriver. On the one hand, personal computers(PCs) already offer sufficiently goodperformance for a large number of users; onthe other, new applications, such asautomatic voice recognition or PC wirelesscommunications, may give further impulsesfor further progress (Compano, 2001).Military applications have a restrictedvolume but are of strategic importance.Thus, it is generally anticipated that mostnew technologies within this area will emergein (US) military use first before eventuallyfinding their way into the civil sector (Saxl,2000). These and other ideas are summarisedin Table 11.

2 . 3 . 3 Pharmaceuticals and medicine

Nanotechnology, combined withbiotechnology, are the underpinningtechnologies pushing the rapid advances in‘genomics, combinatorial chemistry, high

throughput robotic screening, drug discovery,

gene sequencing and bioinformatics and their

applications’ (Saxl, 2000). In medicine,advances can take place at the nanoscalewhere, for example, either passive or activenano-engineered systems can be used that

26

Table 11: Summary of application areas for informatics.

Material/technique A p p l i c a t i o n s Time-scale (to market launch)

Pre 2015

Quantum well structures Telecommunications/optics industry. Quantum well lasers already (pers.comm., Gareth Barry, Potentially very important applications utilised in CD players. Not Imperial College London, in laser development for the data yet optimised for the 22 Nov 2002). communications sector. The aim is to communications market

use fibre optic communications in (i.e. fibre optics): 4–5 years. building and computers. The problems

Quantum dot structures are cost and high temperature operating Quantum dots still in research (source as above). conditions. Quantum well/dot structures stage: 7–8 years.

can potentially solve this problem

Photonic crystal technologies Optical communication sector, i.e. fibre Still in basic R&D, but very(Miles and Jarvis, 2001). optics. Photonic integrated circuits can strong commercial interest

be nearly a million times denser than emerging. electronic ones. Their tighterconfinement and novel dispersionproperties also open up opportunitiesfor very low power devices

Carbon nanotubes in Memory and storage; commercial Commercial prototype nanoelectronics. These hold prototype nanotube-based nanotube-based RAM predictedpromise as basic components (non-volatile); RAM; display in 1–2 years. for nanoelectronics – they can technologies; E-paper.act as conductors, Consumer flat screen bysemiconductors and insulators the end of 2003. ( H o l i s t e r, 2002).

Limited commercialisation of E-paper in 1–2 years.

Spintronics – the utilisation Ultra-high capacity disk drives and A read head has been of electron spin for significantly computer memories. demonstrated that can deal withenhanced or fundamentally storage densities of a terabit per new device functionality square inch. In 2001 Fuji (Science Blog, 2002). announced a new magnetic

coating promising 3-gigabyte floppy disk.

Polymers (Compano, 2001). Display technologies – this sector Some commercialisation, is driven by the electronics consumer e.g. Cambridge Display market. Technologies has been formed

specifically to exploit this t e c h n o l o g y.

Post 2015

Molecular nanoelectronics Circuits based on single molecule and Single atom transistor(including DNA computing) single electron transistors will appear, demonstrated recently. Still(Compano, 2001). initially in special applications. immature, but huge potential

(Miles and Jarvis, 2001).

Quantum information Several researchers have devised Still in pure research phase, processing (QIP) algorithms for problems that are very although some US defence (Compano, 2001). computationally intensive (and thus money has been made available

time-consuming) for existing digital ( H o l i s t e r, 2002). computers, which could be made muchfaster using the physics of quantumcomputers. E.g. factoring large numbers(essential for cryptographic applications),searching large databases, patternmatching, simulation of molecularand quantum phenomena (Anton et al., 2001).


enable the required dose of drug to bedelivered at the correct time to the targetarea, or at the macro-level, such as inducedtissue growth. This reduces unwanted sideeffects, improves patient compliance, leads tolower doses and opens up new possibilities(that would be impossible withoutnanotechnology approaches) (Miles andJarvis, 2001). The size of this market is themain driving force behind such innovation.LaVan and Langer (2001) predict that:‘fundamental changes in drug production

and delivery are expected to affect about half

of the [US]$380 billion world- wide drug

production in the next decade.’ At present,nanotechnology is estimated to have a 1%stake in this, but whole sectors will continueto grow and this contribution is expected toincrease rapidly (Ho, 2002a). The US iswidely recognised as a leader in this area,with a company market share of about 40%,and many applications close to the market.

2.3.3.1 Drug delivery

The most promising aspect ofp h a rmaceuticals and medicine as it relates tonanotechnology is currently drug delivery. Inthe words of LaVan and Langer (2001): ‘It islikely that the pharmaceutical industry will

transition from a paradigm of drug discovery

by screening compounds to the purposeful

engineering of targeted molecules.’ At pre s e n t ,t h e re are 30 main drug delivery products onthe market. The total annual income for all ofthese is approximately US$33 billion with anannual growth of 15% (based on globalp roduct revenue) (Miles and Jarvis, 2001).Two major drivers are primarily re s p o n s i b l efor this increase in the market. First, pre s e n tadvances in diagnostic technology appear tobe outpacing advances in new therapeuticagents. Highly detailed information from apatient is becoming available, thus pro m o t i n gmuch more specific use of pharm a c e u t i c a l s( L a Van and Langer, 2001). Second, theacceptance of new drug formulations isexpensive and slow, taking up to 15 years toobtain accreditation of new drug form u l a s

with no guarantee of success. In re s p o n s e ,some companies are trying to hurry the longclinical phase re q u i red in We s t e rn medicine(Ho, 2002a). However, powerful incentivesremain to investigate new techniques that canm o re effectively deliver or target existingd rugs (Saxl, 2000). In addition, many of thesenew tools will have foundation in curre n ttechniques: a targeted molecule may simplyadd spatial or temporal resolution to anexisting assay. Thus, although many potentialapplications are envisaged, the actual near-f u t u re products are not much more thanbetter re s e a rch tools or aids to diagnosis (Ho,2002a). These are summarised in Table 12.

2 . 3 . 4 E n e r g y

The global energy sector is likely to bep a rticularly affected by coming advances inn a n o t e c h n o l o g y. To illustrate, significantchanges in lighting technologies are expectedin the next 10–15 years. Semiconductors usedin the preparation of light-emitting diodes cani n c reasingly be sculpted on nanoscaledimensions. Projections indicate that suchnanotechnology-based advances have thepotential to reduce world-wide consumptionof energy by more than 10% (NSF, 2001).The various applications showing mostp romise are summarised in Table 13.

Most current photovoltaic (PV) production isbased upon crystalline and amorphoussilicon technologies. However, as Table 13shows, research is now focusing upon newtechnologies which may result in significantreductions in PV costs and/or improvementsin efficiency. Nanotechnology is anticipatedto play an important part in this future.Although total PV power output remainsrelatively low, the industry is growingrapidly-the production of PV modulesexpanded by 40% in 1997 (Saxl, 2000). Thisincrease is largely due to the building andconstruction industry, the largest and fastestgrowing sector at present. In addition,developing countries represent a potentiallyvast market (pers. comm., Jenny Nelson,

28

Table 12: Summary of application areas for nanoscale pharmaceuticals and medicine.

M a t e r i a l / t e c h n i q u e P r o p e rty A p p l i c a t i o n s Time-scale

(to market launch)

D i a g n o s t i c s

Nanosized markers Minute quantities of a E.g. detection of cancer ?i.e. the attachment of substance can be detected, cells to allow earlynanoparticles to molecules down to individual t r e a t m e n t .of interest (Holister, 2002). molecules

‘Lab-on-a-chip’ Miniaturisation and The creation of miniature, Although chips currentlytechnologies (Saxl, 2000). speeding up of the portable diagnostic cost over £1250

analytical process. laboratories for uses in (US$2085) each to the food, pharmaceutical make, within three years and chemical industries; the costs should fall in disease prevention and d r a m a t i c a l l y, making control; and in these tools widely environmental monitoring. a v a i l a b l e .

Quantum dots (pers. Quantum dots can be D i a g n o s i s In early stage of comm., Gareth Barry, tracked very precisely development, but there is Imperial College London, when molecules are ‘bar enough interest here for 22 Nov 2002). coded’ by their unique some commercialisation

light spectrum. . (e.g. Q-dot Inc.).

Drug delivery

Nanoparticles in the Larger particles cannot Cancer treatment. ?range of 50–100 nm enter tumour pores while(Miles and Jarvis, 2001). nanoparticles can easily

move into a tumour.

Nanosizing in the range Low solubility. More effective treatment ?of 100–200 nm (Miles with existing drugs. and Jarvis, 2001).

Polymers (Holister, 2002). These molecules can be Nanobiological drug ?engineered to a high carrying devices.degree of accuracy.

Ligands on a These molecules can be The ligand target receptors ?nanoparticle surface engineered to a high can recognise damaged( H o l i s t e r, 2002). degree of accuracy. tissue, attach to it and

release a therapeutic drug.

N a n o c a p s u l e s Evading body’s immune A Buckyball-based AIDS Early clinical.( H o l i s t e r, 2002). system whilst directing a treatment is just about to

therapeutic agent to the enter clinical trialsdesired site. (Ho, 2002a).

Increased particle Degree of localised drug Slow drug release. ?adhesion (Holister, 2002). retention increased.

Nanoporous materials Evading body’s immune When coupled to sensors, Pre-clinical: an insulin-( H o l i s t e r, 2002). system whilst directing a drug-delivering implants delivery system is being

therapeutic agent to the could be developed. tested in mice.desired site.

‘ P h a r m a c y - o n - a - c h i p ’ Monitor conditions and act E.g. Diabetes treatment. More distant than ‘lab-on-(Saxl, 2000). as an artificial means of a-chip’ technologies.

regulating and maintainingthe body’s own hormonalb a l a n c e .

Sorting biomolecules Nanopores and Gene analysis Current – ?( H o l i s t e r, 2002). membranes are capable and sequencing.

of sorting, for example, left- and right-handedversions of molecules.

Tissue regeneration, growth and repair

N a n o e n g i n e e r e d Increased miniaturisation; Retinal, auditory, spinal Most immediate will be prosthetics increased prosthetic and cranial implants. external tissue grafts; (Miles and Jarvis, 2001). strength and weight dental and bone

reduction; improved replacements; internal b i o c o m p a t i b i l i t y. tissue implants

(Miles and Jarvis, 2001).

Cellular manipulation Manipulation and coercion Persuasion of lost nerve More distant: 5–7 years.(Miles and Jarvis, 2001). of cellular systems. tissue to grow; growth

of body parts.


Table 13: Summary of applications for energy processing.

M a t e r i a l / t e c h n i q u e A p p l i c a t i o n s Time-scale (to market launch)

Power generation (PV technology)

Polymer materials Solar cells (pers. comm., Jenny Nelson, The research stage has( o r g a n i c ) . Imperial College London, 2 Dec 2002). advanced much more quickly

Current developments aim to balance moderate than expected. As a result,efficiency with low cost. Another big advantage p o l y m e r-based PV cells shouldis that these layers can easily be incorporated enter the market in 5 years. into appliances. Current problems stem fromthe material’s instability.

Combinations of Dye-sensitised solar cells made from a thin Low power applications willorganic and hybrid layer (pers. comm., Jenny Nelson, enter market first. Limitedinorganic molecules. Imperial College London, 2 Dec 2002). These commercialisation already

cells are potentially very cheap because occurring (e.g. by Sustainablefabrication is from cheap, low purity materials Technologies International).by simple and low cost procedures (Saxl, 2000). Photocatalytic water treatment.

Quantum wells Quantum-well solar cells (pers. comm., Jenny Pure research.( i n o r g a n i c ) . Nelson, Imperial College London, 2 Dec 2002).

Current research is taking place in high-efficiencyapplications because the infrared part of the solar spectrum may be absorbed.

N a n o r o d s . These structures can be tuned to respond to L o n g - t e r m .different wavelengths of light forming cheapand efficient solar cells (Holister, 2002).

Fuel conversion/storage

Improved fuel Fuel conversion (Saxl, 2000). Current – 3 years.catalysts through nanostructuring.

N a n o t u b e s . Fuel storage. E.g. a methane-based fuel cell 2 years.for powering mobile phones and laptops iscurrently being developed. (Holister, 2002).

Nanoparticles. Vastly increased (e.g. x10) charge and discharge D i s t a n t .battery rate (Holister, 2002).

Imperial College London, 2 Dec 2002). In spite of these developments, however,n a n o t e c h n o l o g y, as a new and radicalt e c h n o l o g y, still faces an uncertain future in this area as a number of altern a t i v etechnologies are also competing for attention(e.g. inorganic silicon). Indeed, it may be 20years before nanotech-based PV beginscompeting as a viable energy source withthis example. As reminiscent of so many ofthe aforementioned applications in thissection, there is much hype but no one re a l l yknows how to achieve these things yet (pers.comm., Jenny Nelson, Imperial CollegeLondon, 2 Dec 2002).

2 . 3 . 5 Defence

Nanoscale informatics, pharmaceuticals andmedicine remain the most high-profile areasof near-term market application. However,Gsponer (2002) contends that the mostsignificant near-term applications ofnanotechnology will be in the militarydomain. This is because micromechanicaland MEMS engineering is historicallyconnected to nuclear weapons laboratories: itwas within this domain that the field ofnanotechnology was born a few decades ago.Today, it is not difficult to understand whynanotechnology might appeal to militaryplanners. Through technologies such as

30

steam navigation, repeating firearms, andhigh explosives, Western powers haveenjoyed virtually unchallenged militarysupremacy throughout the 19th Century(Reynolds, 2002). It is not absurd, then, toimagine that nanotechnology could play asimilar role in the 21st Century. Indeed, newtechnologies, notably IT, are playing anincreasingly important part in modernwarfare-as reflected by recent investments inthe US DoD (see Table 4). Trends such asthese have led leading strategiccommentators, such as David Jeremiah(1995), to conclude that military applicationsof nanotechnology have an even greaterpotential than nuclear weapons to radicallychange the balance of global power in thefuture. Fundamentally, this potential lies in agreater range of military options whendeciding how to respond to aggression. ScottPace (1989) of RAND expands upon this:

‘How might nanotechnology contribute to US

m i l i t a ry power? In peacetime or crisis,

nanocomputers may allow more capable

s u rveillance of potential aggressors. The flood

of data from world-wide sensors could be

culled more efficiently to look for tru l y

t h reatening activities. In low-intensity warf a re ,

intelligent sensors and barrier systems could

isolate or channel guerrilla movements

depending on the local terrain. In

conventional theatre war, nanotechnology

may lead to small, cheap, highly lethal anti-

tank weapons. Such weapons could allow

relatively small numbers of infantry to defeat

assaults by large arm o u red forces. At nuclear

conflict levels, accurate nanocomputer

guidance and low nanomachine pro d u c t i o n

costs would accelerate current trends in the

p roliferation of ‘smart’ munitions. Rather

than requiring nuclear weapons to attack

massive conventional forces or distant, hard

t a rgets, nanotechnology enhancements to

c ruise missiles and ballistic missiles could

allow them to destroy their targets with

conventional explosives. Conventional

explosives themselves might be replaced by

molecular disassemblers that would be rapidly

e ffective, but with less unintended destru c t i o n

to surrounding buildings and populations.’

Other stated applications include (NSF,2001):

• information dominance throughadvanced nanoelectronics

• more sophisticated virtual reality systems

• increased use of enhanced automationand robotics

• required improvements inchemical/biological/nuclear sensing

• design improvements in systems used fornuclear non-proliferation monitoring andmanagement

• combined nanomechanical andmicromechanical devices for control ofnuclear defence systems.

In addition, such nanotechnologies might be‘cleaner’ and ‘safer’ and less likely to causecollateral damage than the technologies theyreplace, making them especially appealing tomilitary planners (Reynolds, 2002). Forexample, MEMS have many potential uses inthe battlefield, largely due to their built-inmechanical functions that allow them to actas sensors and actuators (RAND, 2002).Actuators in particular extend thefunctionality of sensors by allowing them torespond to the environment with the usage offorce. Applications of MEMS in militarysystems include ammunition, petroleum,food, as well as enabling a host of othersmarter, more efficient logistics operations.

The infantry soldier too is anticipated toreceive a nanotech-based ‘makeover’: a newInstitute for Soldier Nanotechnology (ISN)has been created at MIT, with a US Armygrant of US$50 million over five years. The


goal of this research centre is to greatlyenhance the protection and survival of theinfantry soldier using nanoscience (NewScientist, 2002). For example, US armyplanners are hoping to lighten the load thatsoldiers carry into battle (currently around64 kg) by redesigning the equipment fromthe atomic scale up. Current signs indicatethat progress towards these objectives maysoon begin to bear fruit: a Centre forNanoscience Innovation for Defence (CNID)was created in January 2003 to facilitate therapid transition of research innovation in thenanosciences into applications for thedefence sector (Science Blog, 2002). CNID is sponsored by two federal agencies – theDefence Advanced Research Project Agency(DARPA) and Defence MicroElectronicsActivity (DMEA) – to the tune of US$20million over three years.

2 . 3 . 6 Corporate funding

The difficulties involved in drawing uponaccurate corporate data from within thepublic domain are far more substantial than those encountered with re g a rd topublic investment. Thus, a detailed analysisof corporate activity is mainly beyond the scope of this re p o rt. However, it isi m p o rtant to recognise that, urged on by the growing interest (and hype) curre n t l ys u rrounding nanotechnology, spending bybig firms in 2002 is anticipated to match oreven exceed government spending (Holister,2002). Furt h e rm o re, this private investmentis very often at the fore f ront of applicationdevelopment in the marketplace. Helsel(2002) demonstrates this by showing howhistorical funding for technologytransitioning into the US market place is led by corporate sources.

In total, there are an estimated 470 nanotech companies distributed acro s sN o rth America, Europe and Asia (ETCG roup, 2002a). Of these, about 230 arebased in the US, about 130 in Europe, and about 75 in the Asia-Pacific.

2.3.6.1 Transnational companies.

Transnational companies often carry outtheir own nanotech-related R&D. This isbecause they understand thatnanotechnology is likely to disrupt theirc u rrent products and processes, andt h e re f o re recognise the need to understandand control the pace of such implications(DTI, 2002). In this way, some of thew o r l d ’s largest companies, including IBM,M o t o rola, Hewlett Packard, Lucent,Hitachi, Mitsubishi, NEC, Corning, DowChemical and 3M have launched significantnanotech initiatives through their ownv e n t u re capital funds or as a direct result oftheir own R&D (Holister, 2002). In the USand Switzerland for example, IBM isp roviding some US$100 million nanotech-related funding for its hi-tech re s e a rc hlaboratories (DTI, 2002). In Japan too,many of the nation’s largest players havenow entered the nanotech field, includingFuji, Hewlett-Packard Japan, Hitachi,Mitsubishi, NEC and Sony. For example,Toray Industries, a global maker ofsynthetic fibre, textiles and chemicals, isestablishing a US$40 million centrespecialising in nanotechnology andbiotechnology near Tokyo. The building isexpected to be finished by March 2003(Fried, 2002).

Table 14: US historical funding for technology

transitioning into the marketplace.

S o u r c e P e r c e n t a g e

C o r p o r a t e 3 4 %

Federal government 2 9 %

A n g e l s * 2 5 %

State and local government 5 %

Venture capital institutions 4 %

University endowments 3 %

* Angels are individuals who provide capital to oneor more start-up companies. An angel is usuallyaffluent or has a personal stake in the success of theventure. Such investments are characterised by highlevels of risk and a potentially large return oninvestment.

Adapted from Helsel, 2002.

32

2.3.6.2 Start-up companies

At present there are about 100 business start -ups – new business ventures in their earlieststages of development – in operation today,about half of which are located in the US(Thibodeau, 2002). Such companies rely ontheir understanding of where newo p p o rtunities and markets may lie and thusplay an important role in commerc i a l i s i n gre s e a rch. This increase in activity amongsts t a rt-ups is mirro red by the investmentc o m m u n i t y, who, according to Abid Khan(pers. comm., London Centre forN a n o t e c h n o l o g y, 6 Nov 2002), have decidedthat nanotechnology is the ‘next big thing’ –the new computing or biotechnology. Indeed,some large investment groups now havespecialists who follow developments in thesubject. Although such activity tends top roduce little in the way of a cohere n tp i c t u re, business investment innanotechnology start-ups is on the rise. T h e re were over 20 nanotechnologyinvestments in the first half of 2002 in the US and Europe, and more than US$100million invested in the US in the first half of 2002 (Holister, 2002). According toThibodeau (2002), this level of funding isp rojected to increase to US$1 billion by 2003.

2.4 Reality and Hype

2 . 4 . 1 I n t r o d u c t i o n

Nanotechnology advocates have beencriticised within recent years for hyping thepotential impact that nanoscale science andtechnology will have upon the economy andsociety. For example, in response to the NSFclaim that the size of the nanotechnologymarket will reach US$1 trillion in 10 yearstime (Roco and Bainbridge, 2001), TheEconomist (2002) points to ‘nano-

enthusiasts’ being responsible for ‘recklessly

setting impossibly high expectations for the

economic benefits.’ This sentiment is evenechoed by some material scientiststhemselves: Roy (2002) describes the term‘nano’ as a ‘halo regime’ – a term that is sold

to budget managers in order to increasefunding. He concludes that: ‘the [term]

should be new, different, euphonious, and

connected somehow, however tenuously, to

science.’ It is not difficult to identify the kindof claims that can fuel such reaction. Forexample, Pergamit and Peterson (1993) statethat: ‘Humanity will be faced with a

powerful, accelerated social revolution as a

result of nanotechnology. In the near future,

a team of scientists will succeed in

constructing the first nano-sized robot

capable of self-replication. Within a few

short years, and five billion trillion nano-

robots later, virtually all present industrial

processes will be obsolete as well as our

contemporary concept of labor.’

Regardless of the accuracy of these claims,however, there can be no doubt that thelanguage in which they are framed hashelped to attract large amounts ofinvestment. The pinnacle of this came in1997 when the US NNI was launched byPresident Bill Clinton to ‘an extraordinary

amount of hype’ (ETC Group, 2002a).Amongst the various documents produced bythe White House about the subject was oneentitled: National Nanotechnology Initiative:

Leading to the Next Industrial Revolution

(White House Fact Sheet, 2000). The factsheet lists seven ‘potential breakthroughs’

anticipated over the next quarter-century.These include ‘making materials and

products from the bottom-up’ (i.e. bybuilding them up from atoms and molecules)and ‘improving the computer speed and

efficiency of minuscule transistors and

memory chips by factors of millions.’

However, these ambitious claims wereaccompanied with very little seriousinvestigation of their feasibility, or indeedwhether nanotechnology – rather than someother competing technology – will evendeliver within the allotted time-frame.

At present, there is a general understandingamongst industry that the level of hype


s u rrounding nanotechnology has, to someextent, damaged investment potential (DTI,2002). For example, Schultz (2002) advocatesthe need for nanotech re s e a rchers ands u p p o rters to dampen unquestioningenthusiasm for nanotechnology. This isbecause, without discussion of the potentialpitfalls, future nanotechnology re s e a rch couldbe subjected to such extreme pre s s u re thatfunding is jeopardised and re s e a rch pro g re s sis slowed, perhaps halted altogether in somecases. This realisation has quickly lead to anattempt by industry to diffuse some of thewilder claims surrounding the field. GlennF i s h b i n e ’s I n v e s t o r’s Guide to Nanotech and

M i c romachines (2002) provides one suchexample: it is through this type of work thatthe science fiction aspects of the debate arenow receding (pers. comm., Abid Khan,London Centre for Nanotechnology, 6 Nov2002). However, in spite of thesedevelopments, it is clear that the distinctionbetween near- f u t u re applications ofnanotechnology (see Section 2.3) and some of the more visionary aspects of the debatehas become blurred. An attempt todistinguish between the two areas here willhelp draw out some of the more legitimatec o n c e rns currently being voiced aboutnanotechnology in the following section.

2 . 4 . 2 Molecular nanotechnology

The more hyped aspects of nanotechnologyhave generally revolved around MNT.Proponents of this approach suggest thatenvironmentally clean, inexpensive, andefficient manufacturing of structures, devices,and ‘smart’ products, based on the flexiblecontrol of architectures and processes at anatomic or molecular scale of precision, maybe feasible in the near future (i.e. 10–20years from the present). The ambitious goalis to produce complex products on demandusing simple raw materials, such as byinserting the basic chemical elements in amolecular assembly factory to yield acommon household appliance (Nelson andShipbaugh, 1995). These visions have

attracted a great deal of public interest, andimpressive demonstrations have been madeof microscopic devices. For example, inAugust 2001 scientists from OsakaUniversity built the smallest micromechanicalsystem ever, a spring whose arm is only 0.3 µm wide (Ho, 2002a). However,although almost qualifying as a nanodevice,the question of whether it is possible toattain extreme capability and, if so, how todevelop the field, is a point of contention inboth scientific and policy circles (Nelson andShipbaugh, 1995).

In spite of the above controversies, it remainsclear that bottom-up technologies, whilehaving the potential to be immenselyimportant in the longer term, are not likelyin the near future (DTI, 2002). However,some products benefiting from research intomolecular manufacturing may be developedin the near term. As initial nanomachining,novel chemistry and protein engineering (or other biotechnologies) are refined, initialproducts will likely focus on those thatsubstitute for existing high-cost, lower-efficiency products. Likely candidates forthese technologies include a wide variety ofsensor applications, tailored biomedicalproducts (including diagnostics andtherapeutics), extremely capable computingand storage products, and unique, tailoredmaterials (i.e. smart materials usingnanoscale sensors, actuators, and perhapscontroller elements) for aerospace or similarhigh-capability needs (Nelson andShipbaugh, 1995). Predictions of whenbottom-up processes will begin to becomeavailable on a widespread basis vary acrossthe literature. In general, the hyped aspectsof the industry are operating around a 20-year time-scale, with estimates foreconomically viable self-assembly techniquestending to convene around 2015 (Ho,2002a). However, to reach a fully maturenanotechnology society – where it is possibleto manipulate objects on all scales from atomto macroscopic – is expected to take at least

34

35 years (Nelson and Shipbaugh, 1995). Thisis partly due to the economic advantages ofcompeting technologies. For example, withregard to advanced computing, Anton et al.,(2001) state that: ‘the odds-on favourite for

the next 15 years remains traditional digital

electronic computers based on semiconductor

technology. Given the virtual certainty of

continued progress in this area, it is hard to

imagine a scenario in which…quantum-

switch-based computing, molecular

computers, or something else could offer

a significant performance advantage at a

competitive price.’ The major technicalobstacles to development in other areas ofMNT – namely molecular manufacturing,general assembly and nanobots – areexpanded upon below.

2.4.2.1 Molecular manufacturing

To realise molecular manufacturing, a numberof technical accomplishments are necessary(Nelson and Shipbaugh, 1995). First, suitablemolecular building blocks must be found.These building blocks must be physicallydurable, chemically stable, easily manipulated,and (to a certain extent) functionally versatile.The second major area for development is inthe ability to assemble complex stru c t u re sbased on a particular design. A number ofre s e a rchers have been working on diff e re n ta p p roaches to this issue. One uses atomic-f o rce or molecular microscopes with verysmall nanoprobes to move atoms or moleculesa round with the aid of physical or chemicalf o rces. An alternative approach uses lasers toplace molecules in a desired location.Chemical assembly techniques are also beinga d d ressed, including an approach to buildings t ru c t u res one molecular layer at a time. At h i rd major area for development withinmolecular manufacturing is systems designand engineering. Extremely complexmolecular systems at the macroscale willre q u i re substantial subsystem design, overallsystem design, and systems integration, muchlike complex manufactured systems of thep resent day. Although the design issues are

likely to be largely separable at a subsystemslevel, the amount of computation re q u i red fordesign and validation is likely to be quitesubstantial. Perf o rming checks on engineeringconstraints, such as defect tolerance, physicali n t e g r i t y, and chemical stability, will bere q u i red as well (Anton et al,. 2001).

2.4.2.2 Nanobots and other nanoscale devices

This area can be accredited with receivingthe most severe hype, where headline-grabbing predictions include curing cancer,eliminating infections, enhancing ourintelligence, and even making us immortal. In fact, according to Saxl (2000), it will take25 years at least before tiny machinescirculate in the bloodstream cleaning out fatdeposits from our arteries. Indeed, althoughthe implications of such revolutionarytechnologies are awesome, developments thatappear achievable in the short and mediumterm are not particularly dramatic. Perhapsthe most advanced work in this areaconcerns MIT’s BioinstrumentationLaboratory where an autonomous miniaturerobot, dubbed the ‘NanoWalker’, is beingdesigned (MIT, 2002). Measuringapproximately 25 mm2, the nameNanoWalker stems from its ability to takethousands of steps per second in thenanometre range. The ultimate goal of thistype of robotic machine, generically referredto as an assembler, is the construction ofmaterials an atom or molecule at a time byprecisely placing reactive groups. This iscalled ‘positional assembly’ (Holister, 2002).

2.4.2.3 General assembly

The General Assembler is considered to bethe ‘Holy Grail’ of nanotechnology andrepresents the ultimate utility of atom-manipulating nanobots. In general, such anassembling device is regarded as extremelydistant (e.g. more than 25 years). However,there are presently two US companies knownto be going after molecular assembly, inaddition to engineering several ‘magical’assembler dependent products. One of these


companies, Zyvex (2002), aims ‘to become

the leading world-wide supplier of tools,

products, and services that enable adaptable,

affordable, and molecularly precise

manufacturing’ and offers a ‘variety of

products, services, and licensing

opportunities,’ including a number ofnanomanipulation devices. Such nano-advocates claim the first major breakthroughin this area might occur as early as 2007.

2 . 4 . 3 Fundamental barriers to these visions

This report does not intend to refute thatsignificant progress has been made inconstructing macroscale objects using MNTtechniques. Although the building blocks forthese systems currently exist only in isolationat the research stage, it is certainlyreasonable to expect that an integratedcapability could be developed over the next15 years. Such a system might be able toassemble structures with between 100 and10,000 components and total dimensions ofperhaps tens of microns (Anton et al., 2001).In particular, a series of importantbreakthroughs would certainly causeprogress in this area to develop much morerapidly, especially if research continues toaccelerate at today’s rate. However,particularly in light of some of the wilderclaims regarding nanotechnology-enabledfutures, it must also be stressed that,although molecular manufacturing holdssignificant promise, it remains the leastconcrete of all the technologies discussed inthis report. Certainly, there are a number ofmajor technical obstacles to be overcome,some of which might be virtuallyinsurmountable. Indeed, in the most carefullyconsidered dismissal to date, ProfessorSmalley upholds the notion of nanobotreplicators as fundamentally problematic(Smalley, 2001). First, the fingers of suchatomically sized manipulators are too ‘fat’ toallow sufficient control of the reactionchemistry; second, they are too ‘sticky’ – theatoms of the manipulator hands would beadhered to the atom that is being moved.

Furthermore, other commentators such asHo (2002c), point to major problemsconcerning energy sources and dissipation, orjust the sheer complexity of the task at hand.For example, diamond assemblies might berelatively easy to assemble; other structures,such a biological configurations, areinfinitely more complicated.

2.5 Concerns

2 . 5 . 1 I n t r o d u c t i o n

Given the difficulty in fore s e e i n gnanotechnology outcomes and estimatinglikelihood, it is difficult to extrapolatep redictions of specific threats and risks fro mc u rrent trends (Anton et al., 2001). And yet,in spite of this, recent discussions of thepossible dangers posed by future technologies(such as AI, genetic engineering and MNT)have made it clear that analysis of the majorclasses of risks of nanotechnology isw a rranted. Perhaps the greatest difficulty inp redicting the impacts of new technologieshas to do with the fact that, once thetechnical and commercial feasibility of aninnovation is demonstrated, subsequentdevelopments may be as much in the hands ofusers as in those of the innovators (NSF,2001). As a result, new technologies cana ffect society in ways that were not intendedby those who initiated them. Sometimes theseunintended consequences are beneficial, suchas spin-offs with valuable applications infields remote from the original innovation. Agood example of this concerns the early daysof the Internet – the subject is covered in Part2 of this re p o rt. Other times, intendedbenefits may also have unintended or ‘second-o rd e r’ consequences. Intere s t i n g l y, while a fewf a r-sighted scientists are focusing onpotentially negative second-order impacts off u t u re nanotech applications, virtually no onehas been tracking the potentially negativeimpacts of nanotechnology’s pre s e n t - d a yp roducts (ETC Group, 2002a). This section,t h e re f o re, will attempt to distinguish betweenthese two time-frames, as well as intro d u c i n g

36

the main environmental and socio-politicalc o n c e rns. For the purposes of this re p o rt ,‘ l o n g - t e rm’ refers to a hazard that, due tochallenges associated with technologicaldevelopment, is unlikely to manifest itselfwithin a 10–15 year time-frame.

2 . 5 . 2 Environmental concerns

The potential impact of nanostructuredparticles and devices on the environment isperhaps the most high profile ofcontemporary concerns. Quantum dots,nanoparticles, and other throwawaynanodevices may constitute whole newclasses of non-biodegradable pollutants thatscientists have very little understanding of.Essentially, most nanoparticles producedtoday are mini-versions of particles that havebeen produced for a long time. Thus, thelarger (micro) versions have undergonetesting, while their smaller (nano)counterparts have not (ETC Group, 2002a).For example, Vicki Colvin, ExecutiveDirector of Rice University’s Centre forBiological and EnvironmentalNanotechnology (CBEN) has recentlypostulated that nanomaterials provide a largeand active surface for adsorbing smallercontaminants, such as cadmium andorganics. Thus, like naturally occurringcolloids, they could provide an avenue forrapid and long-range transport of waste inunderground water (cited in Colvin, 2002).

2.5.2.1 Infiltrating humans

The concern that nanomaterials could bindto certain common but harmful substancesin the environment, such as pesticides orPCBs, leads to the short - t e rm worry of suchmaterials infiltrating humans. According tothe ETC Group (2002a), at a recent fact-finding meeting at the US Enviro n m e n t a lP rotection Agency (EPA), re s e a rc h e r sre p o rted that nanoparticles can penetrateliving cells and accumulate in animalo rgans. In part i c u l a r, the possibility of toxicelements attaching themselves to otherw i s ebenign nanomaterials inside bacteria and

finding a way into the bloodstream wasacknowledged. In addition, very little workhas been done in order to ascertain thepossible effects of nanomaterials on livingsystems. One possibility is that proteins inthe bloodstream will attach to the surface of nanoparticles, thus changing their shapeand function, and triggering dangero u sunintended consequences, such as bloodclotting. A second possibility relates to theability of nanoparticles to slip past thehuman immune system unnoticed, ap ro p e rty desirable for drug delivery, butw o rrying if potentially harmful substancescan attach to otherwise benignnanomaterials and reside in the body in asimilar manner. According to Colvin (2002),‘it is possible to speculate that nanoscale

i n o rganic matter is generally biologically

i n e rt. However, without hard data that

specifically address the issue of synthetic

nanomaterials, it is impossible to know

what physiological effects will occur, and,

m o re critically, what exposure levels to

re c o m m e n d . ’ To illustrate, this re p o rt showshow nanotubes, should industry pre d i c t i o n sbe realised, are set to become re l a t i v e l yubiquitous within the coming decades –such materials are already finding their wayinto a number of products. But it has notyet been determined what happens if, forexample, large quantities of nanotubes areabsorbed by the human body. Onep rominent concern relates to the stru c t u r a lsimilarities between nanotubes and asbestosf i b res: like the latter, nanotubes fibres arelong, extremely durable, and have thepotential to reside in the lungs for lengthyperiods of time. One recent study,conducted by the National Aeronautics andSpace Administration (NASA), has shownthat breathing in large quantities ofnanotubes can cause damage to lungs.H o w e v e r, as nanotubes are essentiallysimilar to soot, then this is not part i c u l a r l ysurprising (The Economist, 2002). On thewhole, far more experiments are re q u i re db e f o re the issue can be resolved.


2.5.2.2 Self-replication

Self-replication is probably the earliest-recognised and best-known long-term dangerof MNT. This centres upon the idea that self-replicating nanorobots capable of functioningautonomously in the natural environmentcould quickly convert that naturalenvironment (i.e. ‘biomass’) into replicas ofthemselves (i.e. ‘nanomass’) on a globalbasis. Such a scenario is usually referred to as the ‘grey goo’ problem but perhaps moreproperly termed ‘global ecophagy’ (Freitas,2000). The main feature that distinguishesrunaway replication as a long-termenvironmental concern is the extremedifficulty involved in constructing machineswith the adaptability of living organisms. As Freitas (2000) notes:

‘The replicators easiest to build will be

inflexible machines, like automobiles or

industrial ro b o t s … To build a ru n a w a y

replicator that could operate in the wild would

be like building a car that could go off - ro a d

and fuel itself from tree sap. With enough

work, this should be possible, but it will hard l y

happen by accident. Without re p l i c a t i o n ,

accidents would be like those of industry

today: locally harmful, but not catastrophic to

the biosphere. Catastrophic problems seem

m o re likely to arise though deliberate misuse,

such as the use of nanotechnology for military

a g g re s s i o n ’ (see below).

This is not to imply, however, that the riskthat molecular machines designed foreconomic purposes might replicateunchecked and destroy the world should bewritten off altogether: while the dangerseems slight, even a slight risk of such acatastrophe is best avoided (Zyvex, 2002).To this end, David Forrest (1989) hasproduced a set of guidelines to assure thatmolecular machines and their products aredeveloped in a safe and responsible manner.

2 . 5 . 3 Socio-political concerns

C l e a r l y, if scientists are successful in

developing nanofabrication techniques formanufacturing nanoelectronic devices in hugevolumes at very low cost, then the impact onsociety will be enormous. The potentiallyd i s ruptive nature of nanotechnology hasa l ready been highlighted in earlier sectionst h rough its ability to generate major newparadigm shifts in how things are generated,such as a shift from top-down to bottom-upmanufacturing techniques. This section furt h e relaborates upon this and similar concerns.

2.5.3.1 Medical ethics

The ethical questions that have been raised inrecent years following the advancement ofsuch technologies as gene therapy are similarto in scope and philosophy to nanotechnology.For example, the emergence of highly specificd rug therapies, a nanobased technique thatf e a t u res prominently in earlier sections of thisre p o rt, may result in genetic discrimination.That is, discrimination directed against anindividual or family based solely on ana p p a rent or perceived genetic variation fro mthe ‘normal’ human genotype (LaVan andL a n g e r, 2001). The major concern here lies inthe end result of going down such a road: thatthe de-selection of characteristics judgedunwanted by societies (re f e rred to as negativeeugenics) will be viewed as the right,responsible, moral thing to do, as will cure sand enhancements (Wolbring, 2002).S i m i l a r l y, on a longer time-scale, concern sover nanotech applications for enhancing thep e rf o rmance of the human body might alsoarise. A major question here is whether suchenhancements can be forced upon people,either when in a position to make a decisionfor themselves or, more contro v e r s i a l l y,against their will.

2.5.3.2 The nano-divide

If Moore’s law holds and the miniaturisationof PCs continues unchecked well into the21st Century, then it seems likely that, in thelong-term, society will get to a point wherepeople can carry computers 24 hours a day.As Chaudhari (2001) states: ‘We are evolving

38

to the point where every human being will be

connected to any other human or to the vast

network of information sources throughout

the world by a communication system

comprised of wireless and optical fibre

communication links.’ A world in whichinformation is abundant and cheap may wellhave serious privacy implications for thosewho can afford to connect. However, littleconsideration seems to have been given tothose who will clearly not be able to affordto participate. Indeed, many nations arealready witnessing an IT divide, particularlyin reference to Internet usage, that correlateswith inequality in the distribution of wealth.This gap is likely to be exacerbated by anyimpending nanotechnological revolution,forming a so-called ‘nano-divide.’ It isimportant not to underestimate the potentialscale of this: the transition from a pre-nanoto post-nano world could be very traumaticand could exacerbate the problem of havesvs. have-nots. Such differences are likely tobe striking (Smith, 2001).

A quick glance at demographics providessome insight into what such a post-nanoworld might look like. According to theWorld Bank, the Western industrialdemocracies will shrink from 12.7% oftoday’s population to 8.6% by 2025. At thesame time in the developing world thepopulation will double (cited in Jeremiah,1995). The kinds of nanotech-inspiredwonders alluded to throughout this reportmay only be feasible for the 8.6% of the2025 population who live in Westernindustrial democracies, and the upper layerof society in the developing and non-developing world, not for the rural poor andthe underside of all urban populations. Inother words, ‘the differences in the quality of

life will be even starker than today between

these two worlds’ (Jeremiah, 1995). The NSFsupports these sentiments: ‘Those who

participate in the nano revolution stand to

become very wealthy. Those who do not may

find it increasingly difficult to afford the

technological wonders that it engenders.’

(Roco and Bainbridge, 2001). One near-termexample will be in medical care, as nanotech-based treatments may be initially expensiveand hence only accessible to the very rich.

In the longer-term, campaign groups such asthe ETC Group point to what they describeas the ‘corporate concentration’ of ‘material

building blocks and processes that make

everything from dams to DNA.’ This concernarises irrespective of the general doctrine inpatent law that products of nature cannot bepatented because the atomically-engineeredelements of today are able to side-step theissue. For example, C Sixty Inc., a Toronto,Canada-based start-up exercise, has filed aseries of patents, five of which have beengranted, for Buckminsterfullerene. The aimof C Sixty Inc. is to corner the market withrespect to this remarkable molecule and itsvast potential in drug delivery. A big concernof the ETC Group (2002c) is that patentingoffices (such as the US Patent and TrademarkOffice) understand nanotechnology, so thatwhen approached by industry, examinersunderstand what are reasonable boundariesto intellectual property rights.

2.5.3.3 Destructive uses

The potentially catastrophic but long-termdanger that the deliberate misuse ofnanotechnology for military aggression poseshas already been sketched out above. Indeed,Howard (2002) concedes that ‘once the basic

technology is available, it would not be

difficult to adapt it as an instrument of war

or terror.’ Gsponer (2002), on the otherhand, draws attention to the existingpotential of nanotechnology to affectdangerous and destabilising ‘refinements’ ofexisting nuclear weapons designs – suchfourth generation nuclear weapons are newtypes of explosives that can be developed infull compliance with the Comprehensive TestBan Treaty (CTBT). Such developments hintat the worrying possibility of ananotechnology arms race. Zyvex (2002)


sketch out the underlying rationale for suchan occurrence:

‘It is clear that offensive weapons made using

advanced nanotechnology can only be

stopped by defensive systems made using

advanced nanotechnology as well. If one side

has such weapons and the other doesn’t, the

outcome will be swift and very lopsided. This

is just a specific instance of the general rule

that technological superiority plays an

important and often critical role in

determining the victor in battle. Clearly, we

will need much further research into

defensive systems as this technology becomes

more mature.’

2 . 5 . 4 Public acceptance of nanotechnology

In spite of the concerns highlighted above,both pre c a u t i o n a ry principle and industryadvocates agree that there is time to cre a t edialogue and consensus that could prevent thekind of confrontations occurring that plaguedthe development of biotechnology. In thisw a y, the objective of industry is to launchp re-emptive strikes against any problems withpublic acceptance of nanotechnology thatmight arise down the line (Gorman, 2002).The earliest example of this is the Fore s i g h tInstitute, a think-tank founded in 1986primarily to facilitate public understandingand discussion of the policy issuess u rrounding the development and deploymentof nanotechnology. More re c e n t l y, nanotechre s e a rchers have been urged to build on theexample of the Ethical and SocialImplications (ELSI) project (ani n t e rd i s c i p l i n a ry eff o rt within the HumanGenome Project). That is, to ‘take a hard

look at potential ethical and cultural issues,

but follow through much more carefully and

get out ahead of the public’ (Paul Thompson,P rofessor of Ethics at Purdue University,quoted in Leo, 2001). Indeed, the NNI haslong acknowledged a need to integratesocietal studies and dialogues concerning thep e rceived dangers of nanotechnology with itsinvestment strategy, and the resulting White

House Fact Sheet (2000) promised that theimpact nanotechnology has on society fro mlegal, ethical, social, economic, and workforc ep reparation perspectives would be studied.These aims have already been realised tosome extent. For example, the 2001 NSFre p o rt entitled Implications of Nanoscience

and Nanotechnology takes a long, hard lookat a range of hypothetical social ramifications(Roco and Bainbridge, 2001).

This industry strategy has been received withmixed reaction. Some commentators, such asHo (2002a), have praised scientists fori n f o rming the public with ‘clarity and

c a n d o u r.’ Others, on the other hand have notbeen nearly so generous in their assessment.H e rrera (2002), for example, sums up thep resent state of the nanotech industry asbeing comparable to a ‘sitting duck’, just asbiotech was during the 1990s, because it isnot taking the issue of public acceptances e r i o u s l y. Herrera continues: ‘Ask members of

the nanotechnology community if there are

any obvious or potential controversies that

they should be watching for, and they will say

‘no’… Scientists think about ethics but they

d o n ’t let it interf e re with their work.’

At present, the majority of controversy inthis area surrounds the interaction ofnanomaterials with the environment andtheir implications for human health. VickiColvin of CBEN believes that ‘scientists’

experience with other particulate matter

argues for a thorough examination of how

nanoparticles might react in mammalian

systems when they are inhaled or when there

is skin exposure’ (cited in Schultz, 2002). Inaddition, nanotech manufacturing processesneed to be examined for potential healthimpacts, for example the solvents used in thegases produced in the manufacture of carbonnanotubes. Outside of manufacturing,researchers should investigate the possibleconsequences of nanoparticles entering andaccumulating in the food chain. Indeed, someof the ongoing work by CBEN, and other

40

organisations such as NASA and the EPA,has been alluded to above. However, it isbecoming increasingly clear that this workalone is not sufficient for the scope of theseissues. As Colvin (2002) notes:

‘It is critical that more organisations and

people devote time and money to these

questions. This requires a change in the

current climate: of the [US$710 million in

funding for the NNI in the fiscal year 2003,

less than [US$500,000 is devoted to the

study of environmental impact. It is difficult

to convince scientists, or funding managers,

to support environmental impact studies.

The immediate payback for research that

demonstrates ways of using nanomaterials to

cure disease, for example, is greater than the

reward for uncovering that a nanomaterial

may cause disease.’

One way in which prevailing industryattitudes may be influenced is through theidea that information about unintendedeffects (whatever its conclusions), rather than alarming investors, in fact reassures,thus increasing the likelihood that viablenanotechnology products are developed.Most importantly, hard data on theenvironmental effects of nanomaterials couldgo a long way to building the public’s trust(Colvin, 2002). This is in contrast to, forexample, the controversy that surrounded thepesticide DDT in the 1960s and early 1970s:by refusing to acknowledge the demonstrableenvironmental harm caused by DDT, the USchemical industry lost a controversial buteffective product, particularly for control ofmosquitoes and mosquito-borne diseases.

2 . 5 . 5 The regulation debate

The precautionary approach upholds thatregulatory action may be taken, based on thepossibility of significant environmentaldamage, even before there is conclusive,scientific evidence that the damage will occur(European Environment Agency, 2003).Perhaps the most vigorous example of this

concerns the ETC Group, who have calledfor a global moratorium on the manufactureof nanomaterials until such a time when their interactions with living systems aremore fully understood (McCullagh, 2002).Such an appeal is well-placed within thisprecautionary worldview, and nano-advocates have had to respond quickly with a number of forceful counter-arguments. Many of these claims stem fromthe diversity of envisaged nanotechapplications and products (i.e. essentially a vast array of very small components), the difficulties of defining nanotechnology,and its broad interdisciplinary scope. Indeed, the convergence of a wide number of scientific disciplines within the field ofnanotechnology certainly complicates thepracticalities of enforcing such a ban,especially when one considers that pushingresearch underground may increase either the danger of deliberate misuse, or at leastthe difficulty of ensuring that usage remainswithin responsible boundaries.

As an alternative, nano-enthusiasts advocatea more modest regulation structure combinedwith robust civilian research. Such anapproach would focus work on the potentialrisks and benefits of nanotechnology, whilstensuring that safe practices are exported todeveloping countries. (Indeed, it is in theinterests of developing countries to adoptgood practice, otherwise investment willflop). Thus, such a regime should be basedon the monitoring of the sale of suchtechnologies, rather than control. Thissituation is analogous to biotechnology: theDNA experience, for example, suggests thata combination of self-regulation andgovernment co-ordination can answerlegitimate safety concerns while allowingscientific research to flourish (Reynolds,2002). Thus, while there is no way ofknowing, a priori, the unintended and higherorder consequences of nanotechnology, theparticipation of environmental and socialscientists in the field may allow for


important issues to be identified earlier, theright questions to be raised, and necessarycorrective actions to be taken. It does seemlikely that some form of regulatory controlwill be necessary to assure thatnanotechnology is developed safely – ‘safe

designs, safe procedures and methods to test

for potentially hazardous assemblers can be

incorporated into standards by consensus of

interested parties’ (Forrest, 1989). Thegreatest danger, however, appears to beintentional abuse of the technology, socertain aspects of development should beperformed in a secure environment.

2.6 DiscussionWhile many of the nanotechnologies covere din this part of the re p o rt might appearadvanced, it is fair to conclude that mostc o n t e m p o r a ry experimental capabilities inthis area are still in their infancy. This meansthat it is extremely difficult to foresee manyoutcomes that developments in this field willbring over the next 10 years, let alone assesstheir likelihood. Initially, it is probable thatthe impact of nanotechnology will be limitedto a few specific products and services, whereconsumers are willing (or able) to pay ap remium for new or improved perf o rm a n c e .Looking further ahead, controversy surro u n d sthe possibility of realising some of the wildervisions of a nanotech-enabled future. This isin spite of the fact that many of these ideasstem from quite straightforw a rd conceptsfounded in solid science (Holister, 2002);

we are unlikely to witness any radicaldevelopments during the next 15 years unlessa series of fundamental bre a k t h roughs occurbetween now and then. However, as therange of associated tool and fabricationtechniques begin to mature, the field is set tobecome increasingly commonplace in thecoming decades. Ultimately, then, the longer-t e rm structural impact of nanotechnology ona whole range of sectors – in manufacturing,t r a n s p o rt, services and domestic practice –could be substantial in 30–50 years. Thesechanges are likely to be gradual as, on thewhole, the displacement of an old technologyby a new one tends to be both slow andincomplete (NSF, 2001).

In the meantime, a number of well-foundedshort-term concerns remain, many of whichrevolve around issues of human health.Considering past experiences of industryand government mismanagement in this area(notably through GM-related controversy),nano-advocates would do well to sit up andtake note. For, although an externallyimposed nanotech moratorium seems bothunpractical and probably damaging atpresent, industry may find such a fatevirtually self-imposed if they do not take the issue of public acceptance seriously. Thisreport has shown some nano-advocateawareness of environmentally-sound practice.Industry must demonstrate a commitment tothis by funding the relevant research on a fargreater scale than currently witnessed.

42

3.1 Introduction

3.1.1 About AI and robotics

AI has been one of the most controversialdomains of inquiry in computer science sinceit was first proposed in the 1950s. Defined asthe part of computer science concerned withdesigning systems that exhibit thecharacteristics associated with humanintelligence, the field has attractedresearchers because of its ambitious goalsand enormous underlying intellectualchallenges (National Research Council[NRC], 1999). The ultimate aim is to makecomputer programmes that are capable ofsolving problems and achieving goals in theworld as well as humans – the pursuit of so-called ‘strong AI’. This goal has caught theattention of the media, but by no means doall AI researchers view strong AI as worthinvestigating – excessive optimism in the1950s and 1960s concerning strong AI hasgiven way to an appreciation of the extremedifficulty of the problem (Copeland, 2000).To date, progress in this direction has beenmeagre. Because 50 years of failureeventually starts to affect funding, the AIfield has diversified and experts haveestablished themselves in other areas wherethey can be said to have had some success.These new areas are less concerned with thebusiness of making computers think,focusing instead on what can be referred toas ‘weak AI’ – the development of practicaltechnology for modelling aspects of humanbehaviour (Goodwins, 2001). In this way, AIresearch has produced an extensive body ofprinciples, representations, and algorithms.Today, successful AI applications range fromcustom-built expert systems to mass-produced software and consumer electronics.

Robotics, on the other hand, may be thoughtof as ‘the science of extending human motor

capabilities with machines’ (Trevelyan,1999). However, a closer look at thisdefinition creates a more complicatedpicture. For example, a cruise missile,

although not intuitively referred to as arobot, nevertheless incorporates many of thenavigation and control techniques exploredin the context of mobile-robotics research.Furthermore, robots are not necessarilydependent on hardware for their operation.It is possible, for instance, to conceive ofintelligent entities that operate purely withininformation systems – the so-called ‘softbots’or ‘software agents’ – as robots (Doyle andDean, 1996). It is noteworthy, however, thatsuch distinctions between ‘hard’ and ‘soft’are bound to fade in importance in the futureas physical agents enter into electroniccommunication with each other and withonline information sources, and asinformational agents exploit perceptual andmotor mechanisms. It is difficult, then, tostate categorically exactly what constitutes a robot. This report, however, considersrobotics research as the attempt to instilintelligent software with some degree ofmotor capability. Since many of the majorareas of AI research play an essential role inwork on robots, robotics will be consideredhere as a sub-section of AI.

3.1.2 Where are we now?

As alluded to above, the field of AI has notmoved along as quickly as innovators havepredicted. One reason for this has been thedamaging cycle of hype and disappointmentwithin the industry, and the accompanyingrise and fall in research investment4 (pers.comm., Murray Shanahan, Imperial CollegeLondon, 17 Jan 2003). This began in the1960s when general enthusiasm surroundingthe prospects of AI moved in parallel withthe exciting developments of the computer.However, this optimism resulted in downfallduring the 1970s when the work failed toproduce, climaxing in the UK with the highlydamaging 1973 Lighthill Report – agovernment commissioned paper from theScience and Engineering Research Councilwhich damned AI and recommendedwithdrawal of research funding. In addition,the same kind of official doubts which the

3. Artificial Intelligenceand Robotics


Lighthill Report made explicit in the UK lay,less explicitly, behind a similar slow down inresearch funding in the US (Malcolm, 2001).The next big rise in AI funding occurred inthe 1980s, mainly in reaction to Japaneseenthusiasm for the field. In Japan, the 5thGeneration project was born; the UK reactedthrough the Alvey initiative, which nowfocused on ‘knowledge based systems’ so asto avoid any awkward parallels betweencurrent research and the previouslycondemned AI. Again, both projects werecharacterised by a lack of progress and AIresearch failed to make it into themainstream. Most recently, the early to mid1990s has seen the emergence of softwareagents, and the resulting excitement has onceagain sparked a rise in investment. Inaddition, the field of robotics has becomemuch more influential of late, particularlythrough the entertainment industry. Many of these developments are described in moredetail later on in this report.

Today, AI is about at the same place the PCindustry was in 1978 (Brooks, 2001) – thewaves of enthusiasm that accompanied thedevelopments of computers have long goneand researchers are beginning to come toterms with how hard the problems of AIreally are. However, technological know-howis not the only obstacle that the AI industryfaces – another is the purported ‘AI effect’whereby the existence of AI in modernsoftware products go largely unnoticeddespite the widespread use of suchapplications (Stottler Henke, 2002). Indeed,AI is considered by some researchers to be anunimplementable technology: as soon as thetechnology advances, the perspective shifts,and the quality of intelligence passes to thoseactivities that are still only in the humandomain (Joseph, 2001). For example, manyof those in industry do not use the term‘artificial intelligence’ even when theircompany’s products rely on some AItechniques (Stottler Henke, 2002). The exactreasons for the AI effect are uncertain, but

it is likely that the phenomenon developed inreaction to the kind of historical tendenciesto oversell the industry alluded to earlier.

3.2 Aspects of Research

3 . 2 . 1 I n t r o d u c t i o n

The above section has described howresearchers have re-evaluated theirexpectations with regard to achieving strongAI. Associated with this reality check is therecognition that classical attempts atmodelling AI, based upon the capabilities ofdigital computers to manipulate symbols, areprobably not sufficient to achieve anythingresembling true intelligence. This is becausesymbolic AI systems, as they are known, aredesigned and programmed rather thantrained or evolved. As a consequence, theyfunction under rules and, as such, tend to bevery fragile, rarely proving effective outsideof their assigned domain (Hsuing, 2002). Inother words, symbolic AI is proving to beonly as smart as the programmer who haswritten the programmes in the first place.

In realisation of this, scientists are beginningto look much more closely at the mechanismsof the brain and the way it learns, evolvesand develops intelligence from a sense ofbeing conscious (Aleksander, 2002). Forexample, AI software designers are beginningto team up with cognitive psychologists anduse cognitive science concepts. Anotherexample centres upon the work of the‘connectionists’ who draw attention tocomputer arc h i t e c t u re, arguing that thea rrangement of most symbolic AIp rogrammes is fundamentally incapable of exhibiting the essential characteristics of intelligence to any useful degree. As ana l t e rnative, connectionists aim to develop AIt h rough artificial neural networks (ANNs).Based on the stru c t u re of the nervous system,these ‘computational-cognitive models’ aredesigned to exhibit some form of learning and ‘common-sense’ by drawing linksbetween meanings (Hsiung, 2002). ANNs,

44

then, work in a similar fashion to the brain:as information comes in, connections amongp rocessing nodes are either strengthened (ifthe new evidence is consistent) or weakened(if the link seems false) (Khan, 2002).

The emergence of ANNs reflects anunderlying paradigm change within the AIresearch community and, as a result, suchsystems have undeniably received muchattention of late. However, regardless of theirsuccess in creating interest, the fact remainsthat ANNs have not nearly been able toreplace symbolic AI. As Grosz and Davis(1994) remark: ‘[Symbolic AI has] produced

the technology that underlies the few

thousand knowledge-based expert systems

used in industry today.’ A major challengefor the next decade, then, is to significantlyextend this foundation to make possible newkinds of high-impact application systems. Asecond major challenge will be to ensure thatAI continues to integrate with related areasof computing research and other fields(Doyle and Dean, 1996). For example, thekinds of developments described in Section 2for nanotechnology may go some way toaccelerating progress in AI, particularlythrough the sensor interface. For thesereasons, the list of main research areas thatfollows should be regarded as neitherexhaustive nor clear-cut. Indeed, futurecategorisations will again change as the fieldsolves problems and identifies new ones.

3 . 2 . 2 L e a r n i n g

According to Daniel Weld (1995) of theUniversity of Washington, machine learningaddresses two interrelated problems: ‘the

development of software that improves

automatically through experience and the

extraction of rules from a large volume of

specific data.’ Systems capable of exhibitingsuch characteristics are important becausethey have the potential to reach higher levelsof performance than systems that must bemodified manually to deal with situationstheir designers did not anticipate (Grosz and

Davis, 1994). This, in turn, allows softwareto automatically adapt to new or changingusers and runtime environments, and toaccommodate for the rapidly increasingquantities of diverse data available today.When designing programmes to tackle theseproblems, AI researchers have a variety oflearning methods at their disposal. However,as alluded to above, ANNs represent one ofthe most promising of these.

3.2.2.1 Artificial neural networks

There are many advantages of ANNs andadvances in this field will increase theirpopularity. Their main value over symbolicAI systems lies in the fact that they aretrained rather than programmed: they learnto evolve to their environment, beyond thecare and attention of their creator (Hsuing,2002). Other major advantages of ANNs liein their ability to classify and recognisepatterns and to handle abnormal input data,a characteristic very important for systemsthat handle a wide range of data.Furthermore, many neural networks arebiologically plausible, which means they mayprovide clues as to how the brain works asthey progress. Like the brain, the power ofANNs lies in their ability to processinformation in a parallel fashion (that is,process multiple chunks of datasimultaneously). This, however, is where thelimitations of such systems begin to arise:unfortunately, machines today are serial –they only execute one instruction at a time.As a consequence, modelling parallelprocessing on serial machines can be a verytime-consuming process (Matthews, 2000a).A second problem relates to the fact that it isvery difficult to understand their internalreasoning processes and therefore to obtainan explanation for any particular conclusion.As a result, they are best used when theresults of a model are more important thanunderstanding how the model works. To thisend, these systems are often used in stockmarket analysis, fingerprint identification,character recognition, speech recognition,


and scientific analysis of data (StottlerHenke, 2002).

3 . 2 . 3 Reasoning about plans,

programs and action

Intelligent systems must be able to plan – todetermine appropriate actions for theirperceived situation, and then execute themand monitor the results. However, in spite ofthe fact that this area has been under activeresearch since the 1950s, AI planningapplications are furthest from human-level(Grosz and Davis, 1994). Ordinary people,for example, manage to accomplish anextraordinary number of complex tasks justusing simple, informal thought processesbased on a large amount of commonknowledge. AI, on the other hand, is farbehind humans in using such reasoningexcept for limited jobs, and tasks that relyheavily on common-sense reasoning areusually poor candidates for AI applications(Stottler Henke, 2002). In the past,researchers have mainly had to rely on thedevelopment of algorithms that‘automatically construct and execute

sequences of primitive commands in order to

achieve high-level goals’ (Weld, 1995). Morerecently, the field of plausible reasoning hasdemonstrated its feasibility in tackling theproblem of representing, understanding, andcontrolling the behaviour of agents or othersystems in the context of incomplete orincorrect information (Weld, 1995). Anotherdevelopment that may lead to significantadvances in the area of artificial reasoning isfuzzy logic. Traditional Western logic systemsassume that things are either in one categoryor another. Yet in everyday life, we know thisis often not precisely so. Fuzzy logic, then,provides a way of taking into account ourcommon-sense knowledge that most thingsare a matter of degree when a computer isautomatically making a decision (StottlerHenke, 2002). Thus, in spite of thedifficulties inherent in this field of AI,planning systems have been successfullydeveloped for several tasks to date, including

factory automation, military transportationscheduling, and medical treatment planning.These will be covered in more detail below.

3 . 2 . 4 Logical AI

This type of reasoning concerns what aprogramme knows about the world ingeneral, the facts of the specific situation inwhich it must act, and the goals that it mustaccomplish (Grosz and Davis, 1994). Suchconcepts are held within the programme inthe form of sentences of some mathematicallogical language. The most successfulexample of this is an expert system, createdwhen a ‘knowledge engineer’ interviewsexperts in a certain domain and tries toembody their knowledge in a computerprogramme for carrying out some task, suchas diagnosis. However, the usefulness ofcurrent expert systems also depends on theirusers demonstrating a certain level ofcommon-sense too.

3.2.4.1 Algorithms and genetic programming

An algorithm is defined as a ‘detailed

sequence of actions to perform to accomplish

some task’ (FOLDOC, 2003). One branch ofalgorithm theory, genetic programming, iscurrently receiving much attention. This is atechnique for getting software to solve a taskby ‘mating’ random programmes andselecting the fittest in millions of generations.Khan (2002) elaborates: ‘Genetic algorithms

use natural selection, mutating and

crossbreeding within a pool of sub-optimal

scenarios. Better solutions live and worse

ones die – allowing the programme to

discover the best option without trying every

possible combination along the way.’

3 . 2 . 5 C o l l a b o r a t i o n

The ubiquity of computers, networks anddistributed information resources means thatcollaboration between these entities isimportant. The field of multiagent co-ordination concerns itself with the problemof endowing agents with the ability tocommunicate with each other to reach

46

mutually beneficial agreements (Grosz andDavis, 1994). In addition, specialisedtechniques must also be developed thatenable an agent to represent and reasonabout the capabilities of other agents (Weld,1995). These types of systems are dealt withby the EU Disappearing Computer projectand are expanded upon later (Section3.3.5.1) due to their focus on spatiallydistributed artefacts.

3 . 2 . 6 P e r c e p t i o n

Many AI systems require an ability to handle several different types of perceptualinformation (Grosz and Davis, 1994). The most important of these are expandedupon below.

3.2.6.1 Pattern recognition

The speed with which people extractinformation from images makes vision thepreferred perceptual modality for mostpeople in the majority of tasks, thus implyingthat easy-to-use computers should be capableof both understanding and synthesisingimages. One of the goals of computer-visionresearch is image understanding andclassification. Depending on the application,the imagery to be understood might include ascanned document page, a mug shot, anaerial photograph, or a video of a home oroffice scene (Weld, 1995). Typical state-of-the-art tasks include facial recognition; objectrecognition and reconstruction; handtracking and gesture recognition; anddocument analysis and recognition. However,while current computer-vision techniques arecapable of impressive feats under controlledconditions, such techniques often prove to bebrittle and non-robust under real-worldconditions (Grosz and Davis, 1994).

3.2.6.2 Understanding natural language

The ultimate goal of natural language-processing research is to create systems ableto communicate with people in naturallanguages. Such communication requires anability to understand the meaning and

purpose of communicative actions, such asspoken utterances, written texts, and thegestures that accompany them and an abilityto produce such communicative actionsappropriately. These abilities, in their mostgeneral form, are ‘far beyond currentscientific understanding and computingtechnology’ (Weld, 1995). However, thepotential relevance of natural languageprocessing to industry is immense, as suchsystems could be central to the nextgeneration of intelligent interface.

3.2.7 Human–computer interaction

This area of AI follows on from perceptionin that people use a number of differentmedia to communicate, including: spoken,signed and written languages; gestures;sounds; drawings; diagrams; and maps(Grosz and Davis, 1994). In particular,knowledge representation is important due toits powerful effect on the prospects for acomputer or person to draw conclusions ormake inferences from that information(Stottler Henke, 2002). Consequently, workin this area seeks to discover expressive,convenient, efficient, and appropriatemethods for representing information aboutall aspects of the world.

3.2.8 Public funding

A c c o rding to hi-tech consultancy, Gart n e rDataquest (cited in BBC, 2002), one billionPCs have been sold across the world, withnumbers anticipated to rise rapidly in thenext few years, reaching the two billionmark in by 2008. The level ofi n t e rconnectedness between such machinesis also set to rise: in this decade, half abillion human-operated machines andcountless computers – in the form ofappliances, sensors, controllers, and the like– will be linked (Dertouzos, 1999). This, int u rn, will lead to an explosion in theI n t e rnet economy. To d a y, some US$50billion changes hands over this system, butby 2030 this flow will amount to US$4trillion of today’s dollars, or one quarter of


the world’s economy. Obviously, in such a future scenario, the extraord i n a r i l ysophisticated systems used to contro lcommunications, power, stock exchanges,and monetary assets can break down andmight come under attack. To d a y, given therelative complexity and unreliability of theI n t e rnet, it is not surprising thatcommentators view this scenario withi n c reasing trepidation. AI, then, is seen bymany as having an essential role in a futurew h e re commercial and military inform a t i o nw a rf a re is a major, perhaps dominant,characteristic. In addition, AI is touted bymany (e.g. see Dertouzos, 1999) as havingthe potential to greatly improve humanp roductivity and ease of use within thisp rospective network.

As computer science, and AI in part i c u l a r, is considered to be of strategic import a n c e ,it is worth here briefly examiningg o v e rnment funding in this area. In general,computer science receives a relatively smallp ro p o rtion of the re s e a rch funding in manycountries, even if anecdotal evidencesuggests that the fractions are incre a s i n g(Schneider and Robb, 2001). This is in spiteof the fact that public funding has played ani m p o rtant part in AI re s e a rch in the past,l a rgely because of the field’s high-riskconceptual challenges. However, the pictureis complicated by the fact that intern a t i o n a lcomparisons of re s e a rch funding in this are aa re difficult to make, since diff e re n tcountries use diff e rent funding methods.France and Japan, for example, rely heavilyon national re s e a rch institutes andlaboratories, rather than expecting mostre s e a rch to be done in universityd e p a rtments. In addition, funding for AIre s e a rch is re p o rted far less thoroughly thanit is for nanotechnology. Consequently,relevant information has often beeno b s c u re, and it has been necessary insteadto re p o rt funding in computer science ingeneral. As a rule, AI budgets will re p re s e n ta small pro p o rtion of these figure s .

3.2.8.1 The US

Historically in the US, the concept of AIoriginated in the private sector, but thegrowth of the field has depended largely onpublic investments. Today, computer scienceresearch in the US is funded by a number ofgovernmental agencies. Total US governmentcomputer science research expenditures in1998 were US$1399 million, withapproximately one third devoted to whatwas described as ‘basic research’ (Schneiderand Robb, 2001). Three agencies (NSF,DARPA, and the Department of Energy[DOE]) in the US together support US$365million of this work, and the NSF isresponsible for funding the lion’s share(Schneider and Robb, 2001). In addition, anumber of other agencies are of note. Theseinclude the National Institutes of Health(NIH) and NASA which have also pursuedAI applications of particular relevance totheir own separate agendas (NRC, 1999).

Of these institutions, DARPA is credited withconsiderable advancement of the field fromthe 1960s onwards. This hardly comes assurprising when one considers the close linkthat exists between the military andcomputer science – in fact, the earlydevelopment of computing was virtuallyexclusively limited to military purposes(Matthews, 2000b). The most famousexample of this concerns the development ofthe Internet, in which DARPA played acentral role in the 1970s and 1980s5. Morerecently, less visible but arguably equallysignificant developments have come to thefore. For example, a 1994 report by theAAAI paraphrased a former director ofDARPA, saying that DART (the intelligentsystem used for troop and materialdeployment for Operation Desert Shield andOperation Desert Storm in 1990 and 1991)‘justified DARPA’s entire investment in AI

technology’ (cited in NRC, 1999). Oneconsequence of this is that modern-day battlerelies heavily on data networks. This hasbeen stressed by A. Michael Andrews, the US

48

Army’s Deputy Assistant Secretary forResearch and Technology, saying:‘Everything relies on a reliable and secure

network. Without it, our vulnerability is

exposed’ (quoted in Machan, 2002).

To d a y, DARPA’s funding for AI re s e a rch iss p read among a number of pro g r a m m ea reas, each with a specific application focus.For example, funding for AI is included inthe Intelligent Systems and Softwarep rogramme, which received roughly US$60million in 1995. This applied re s e a rc hp rogramme is intended to leverage work inintelligent systems and software thats u p p o rts military objectives, enablingi n f o rmation systems to assist in decision-making tasks in stressful, time-sensitivesituations. Additional DARPA funding forAI is contained in the Intelligent Integrationof Information programme, which isintended to improve commanders’ aware n e s sof battlefield conditions. DARPA continuesto fund some of the more basic re s e a rch inAI as well. Such funding is included in itsi n f o rmation sciences budget, which declinedf rom US$35 million to US$22 millionannually between 1991 and 1996. The AIfunding supports work in softwaretechnology development, human-computeri n t e rfaces, micro e l e c t ronics, and speechrecognition and understanding (NRC, 1999).

In addition to DARPA, NASA has also builtup a reputation for high-risk, high-impact AIre s e a rch. One obvious development is thePathfinder robot, which used a number ofm o d e rn AI and robotics techniques to explorethe surface of Mars. Another is the DeepSpace One mission in which an ‘autonomous’c o n t roller (i.e. without human interv e n t i o n )was able to fly a spacecraft for part of amission and exceeded all perf o rmance goals.NASA is expected to continue exploring thistechnology heavily into the future and, basedon these earlier successes, can be considere das a major innovative player within this field( H e n d l e r, 2000).

3.2.8.2 Japan and Europe

Although the US has played a central role indeveloping the AI re s e a rch agenda, othercountries and regions have also played theirp a rt. One of the most notable examples ofthis occurred in the early 1980s when bothJapan and Europe dramatically incre a s e dtheir funding of AI re s e a rch, partly as areaction to the newly emerged expert systemsi n d u s t ry. One of the most ambitious pro j e c t su n d e rtaken was the 5th GenerationComputer Systems Project, an attempt tocombine European ingenuity with Japaneseindustrial skill in order to develop a new sortof AI that might rival the US’s domination inthe field. However, 5th Generation pro j e c ttechnology never really made it into them a i n s t ream, largely because its inflexiblet h e o retical basis was found to be inferior tothe less elaborate, ro u g h - a n d - re a d ya p p roaches to AI development pursued by theUS (Joseph, 2001). The latest collaborativeattempt by these two parties to break UShegemony in AI is the Real World ComputingP roject (RWCP), or the 6th GenerationComputing Project, a 10-year pro g r a m m ethat started in 1992 (around the end of the5th Generation project). This time the RW C Phas a much broader remit: to focus on avariety of diff e rent ‘softer’ technologies thatuse neural or fuzzy techniques. Thus, there s e a rch components are much more spre a dout, and appear to have been selected with aneye for more practical applications of thelatest technology (Joseph, 2001).

In addition to the combined effort above,both parties have also more recentlyestablished research programmes of theirown. In Japan, for example, the NationalInstitute of Informatics (NII), an inter-university research institute under theMinistry of Education, Culture, Sports,Science and Technology (MEXT), is pursuinga programme to expand the field of IT.Established in April 2000, intelligent systems,which form one component of MEXT’sseven-sided agenda, aim to develop advanced


technology for next-generation symbioticrobots and systems, and new models forinformation sharing and exchanging (NII,2002). This programme is closely linked tothe Japanese government’s seven-year plan todevelop humanoid robots. In fact, Japan isthe clear leader in using industrial robots asit accounts for over half of all units in theworld (The Economist, 2001).

The EU, too, is engaged in AI-re l a t e dre s e a rch. Perhaps the most ambitious of thisis related to the EU Framework VI pro p o s a lfor spending €16.29 billion over 2002–2006,of which 27% is destined for IT (Schneiderand Robb, 2001). In addition to central EUfunding, individual European states are alsodeveloping their own re s e a rch agendas. In theUK, the Information Te c h n o l o g y / C o m p u t e rScience (IT/CS) Programme in the Engineeringand Physical Sciences Research Council(EPSRC) budget for 2000/2001 is £70.3million; investment in computer sciencere s e a rch is about 45% of this (EPSRC, 2003).Other EU countries are also of intere s t ,p a rticularly Germ a n y, France andScandinavia, where the latter is part i c u l a r l ywell advanced in the use of computing andI T. However, other smaller countries are notmaking much in the way of substantialcommitments to computer science re s e a rch.

3.3 Applications

3 . 3 . 1 I n t r o d u c t i o n

The above section has demonstrated thediverse and multifaceted nature of AIresearch, and this work has resulted in anextensive body of principles, representations,algorithms, and spin-off technologies (Weld,1995). The relative state of infancy ofresearch into strong AI means that this fieldcan be put aside for the time being. Rather,this section will attempt to elaborate uponweak applications of AI, where, it is fair tosay, considerable effort in this area hasresulted in some real-world product success.In fact, the actual and potential uses of weak

AI are virtually endless: one measure of thegrowth of practical applications is thenumber of patents mentioning the term AIand related terms. According to the USPatent Office, only about 100 patentsspecifically mentioned AI a decade ago; incontrast, in 1999 about 1,700 patentsmentioned AI with another 3,900 or somentioning related terms (Buchanan andUthurusamy, 1999). However, it is worthbearing in mind that the actual prevalence ofemerging AI technology may be greater thanthis due to classification-related difficultiesand the fact that such products are morelikely to be embedded in some larger systemthan a stand-alone machine. In general, suchapplications are used to increase theproductivity of knowledge workers byintelligently automating their tasks, or tomake technical products of all kinds easier touse for both workers and consumers throughintelligent automation of their complexfunctions (Stottler Henke, 2002). It ispossible now to identify four families ofintelligent systems that have broadapplicability across a wide range of sectors(Grosz and Davis, 1994). These areintelligent simulation systems; intelligentinformation resources; intelligent projectcoaches; and robotics.

3 . 3 . 2 Intelligent simulation systems

These applications are commonly used in anumber of different scenarios. First, anIntelligent Simulation System (ISS) may begenerated to learn more about the behaviourof an original system, when the originalsystem is not available for manipulation. Themodelling of climate systems is a goodexample. Second, the original system maynot be available because of cost or safetyreasons, or it may not be built yet and thepurpose of learning about it is to design itbetter (Stottler Henke, 2002). Third, an ISSmight be employed for training purposes inanticipation of dangerous situations, whenthe cost of real-world training is prohibitive.Such technologies are particularly well-

50

advanced in military applications through thesimulation of war ‘games’. Another very bigbusiness in the realm of ISSs is the video-game market, comparable to the filmbusiness in size. AI systems have becomefundamental to this industry because, unlikein film, it is often up to a computer or gameconsole to create a sense of reality for thegame-player. Such standards of realism aregoing up all the time (Broersma, 2001).

3.3.3 Intelligent information resources

Intelligent systems must be able to provideaccess to a wide variety of information,including visual and audio data, in additionto commonplace structured databases (Groszand Davis, 1994). One development in thisarea that is receiving much attention is ‘datamining’, the extraction of general regularitiesfrom online data (Weld, 1995). This area isbecoming increasingly important due to thefact that all types of commercial andgovernment institutions are now logginghuge volumes of data and require the meansto optimise the use of these vast resources(Stottler Henke, 2002). Indeed, according to the market research firm IDC (cited inDalesio, 2002), revenue from sales of alltypes of data mining software are anticipatedto grow from about US$540 million this yearto about US$1.5 billion in 2005.

Looking beyond data mining, othertechnologies are also appearing on thehorizon. For example, SilverEggTechnologies, a Japanese venture company,have developed Aigent, a system thatobserves which product categories acustomer clicks on, and then makesintelligent guesses about that customer’spreferences (Joseph, 2001). Anotherdevelopment in this area concerns the‘heuristic’: ‘A rule of thumb, simplification,

or educated guess that reduces or limits the

search for solutions in domains that are

difficult and poorly understood’ (FOLDOC,2003). Thus, in terms of AI, heuristics is away of trying to discover something or an

idea embedded in a programme. By 2006, itis anticipated that companies will be able touse this kind of software to analyse customerfeedback, whether it comes from the Internet,call centres, or sidewalk surveys. Marketresearch divisions, too, will be able to bettertrack competitors, sales trends and researchextracted from huge volumes of patents,scientific articles and news reports (Dalesio,2002). These developments hint of the ‘nextbig thing’ in industry – ‘business intelligence’.These systems, already in limited applicationtoday, improve on data mining services bypresenting their findings in more usefulformats – using advanced visualisation tools– and by deploying AI to look for patternsthat human users might not look for. Thepotential value of such technology tobusiness has already created fiercecompetition: established software companieslike IBM, Microsoft and Oracle, along withyounger competitors like Business Objects,MicroStrategy and Moreover.com, are vyingfor their share of a market that is expected togrow from US$3.5 billion in 2002 to US$8.8 billion in 2004 (Miller, 2001).

In general, the above examples carry outtasks for one Web site or organisation.However, some innovators envisage thetechnology going a lot further than this. For instance, it is not hard to imagine afuture world of semi-autonomous agents,roaming the Web and carrying out varioustasks for their owners. Such agents could begiven a rough idea of what we want, dosome comparison-shopping, and order thebest deal, just like a real personal assistant.Ultimately, virtual organisations composed of autonomous agents, which could formspontaneously to carry out a specific taskand then disband again, might be possible(Broersma, 2001).

3 . 3 . 4 Intelligent project coaches

This section re p resents the most diverserange of applications: intelligent pro j e c tcoaches can function as co-workers, assisting


and collaborating in a wide range of design or operations teams for complexsystems. For basic personal use, ‘interf a c eagents’ are computer programs that employ AI techniques to provide activeassistance to a user during computer- b a s e dtasks. These agents acquire their competenceby learning from the user as well as fro magents assisting other users. To date, several prototype agents have been builtusing this technique (Maes, 1994). Forexample, US start-ups, such as Saff ro nTechnology and Manna, are marketings o f t w a re tools that learn the individualu s e r’s buying patterns and makepersonalised recommendations accord i n g l y.

In addition to interface agents, the next 10 years are likely to see rapid AIdevelopment occurring in speech re c o g n i t i o n( H e n d l e r, 2000). Indeed, computer speechinput has already arrived and isc o m m e rcially available – many telephones e rvices use speech recognition at present. In addition, cell phones without keypads arelikely to reach the market as early as nexty e a r. These devices are anticipated toenhance the use and appeal of the mobileI n t e rnet by allowing users to call up anyWeb page from a mobile device just byspeaking its address. Voice recognition alsohas security applications: in a demo at the3GSM World Congress in Febru a ry 2002,Mitsubishi Electric demonstrated a SIM cardfeaturing voice validation softwaredeveloped by Domain Dynamics Ltd. Thes o f t w a re provides a ‘biometric template’ thatcan recognise a person’s voice to pro p e r l yidentify a user – ‘a necessity when pro v i d i n g

access to corporate or private databases over

the Intern e t ’ ( M o k h o ff, 2002). With re g a rdto speech re c o g n i t i o n ’s natural successor,natural language processing, suchtechnology is, to date, poorly developed andcomputers are not yet able to even appro a c hthe ability of humans to extract meaningf rom natural languages (Stottler Henke,2002). However, due to the many potentially

valuable practical applications of thist e c h n o l o g y, developments in this area areexpected to advance quickly. For example,automated language translation also looksset to mature sometime between 10–15 years from present.

P e rhaps the most ambitious examples of AI development that are currently occurr i n gin this area relate to computer learning. One example is the ANN, Falcon. Designedby San Diego-based HNC Software, Falconmaintains a profile of how, when, and wherecustomers use their credit cards and, fro mthis, develops an ability to discern ‘deviant’b e h a v i o u r. To date, this system is used bynine of the ten leading US credit cardcompanies: they claim it has improved fraud detection rates from 30–70% (Khan,2002). Another example – and one that isp robably the most challenging in ANNdevelopment today – is being undertaken by DARPA, who have launched an initiativeto develop a cognitive (i.e. thinking) system.The aim of this system is to reason in avariety of ways, learn from experience, and adapt to surprises. In the words ofMelymuka (2002): ‘It will be aware of its

behaviour and explain itself…It will be able

to anticipate diff e rent scenarios and pre d i c t

and plan for novel futures.’ The ultimateaim is to develop cognitive systems capableof assisting or replacing soldiers onh a z a rdous duty or civilians responding to toxic spills or disasters.

In addition to AIs that focus on novel waysof learning, other programmes exist whichcan be said to primarily reason. Perhaps themost successful example in operation todayis the Smart A i r p o rt Operations Centre, a logistics programme created by AscentTe c h n o l o g y. This AI uses genetic algorithmsto plan airport timetables by calculatinghow to optimise complicated scenarios.Other reasoning programmes are based onheuristic classification – a form of expertsystem – and are generally considered the

52

most feasible given the present knowledge ofAI. These AIs have found their way intocockpits of fighter-pilots, where their mainrole is to reduce the workload on the pilotby providing advice in certain stre s s f u lsituations (Matthews, 2000b).

3.3.5 Robotics

A distinction has already been drawn above (Section 3.1.1) between ro b o t sworking in informational environments and robots with physical abilities. Oneadvantage of the former is that there islittle need for investment in additionalexpensive or unreliable robotic hard w a re as existing computer systems and networksp rovide adequate sensor and eff e c t o re n v i ronments. On the other hand, the kinds of robotics systems elaborated onh e re, physical robots, re q u i remechanisation of various physical sensoryand motor abilities (Doyle and Dean,1996). The challenges involved inp roviding such a latter environment areconsiderable, especially when completeautomation is sought, as in Honda’shumanoid ASIMO pro j e c t6. Thus, ratherthan focus on the ambitious and distantgoal of relative autonomy, this re p o rt picksup on Trevelyan (1999) who points outthat complete automation is oftenunfeasible, impossible, or simply unwanted.Indeed, much of today’s robotics re s e a rc hfocuses instead on far humbler goals, suchas simplicity, force control, calibration anda c c u r a c y. Thus, we can see that, to someextent, the field of robotics has followedsimilar lines as that of AI, attempting torebound from the overly optimisticp redictions of the 1950s and 1960s, andcoming up against more contemporaryp roblems not dissimilar to the AI eff e c t .Indeed, while few of the innovations thate m e rge from the work of ro b o t i c sre s e a rchers ever appear in the form ofrobots, or even parts of robots, their re s u l t sa re widely applied in industrial machinesnot defined as so (Trevelyan, 1999).

In spite of these significant challenges, thereare some good examples of AI-controlledrobotic systems. For instance, TriPathImaging has built FocalPoint, a diagnosisexpert system that examines Pap smears forsigns of cervical cancer. FocalPoint screensfive million slides each year, or about 10% ofall slides taken in the US and, like human labtechnicians in training, teaches itself bypractising on slides that pathologists havealready diagnosed. Thus, one big advantageof such a system is that, if implementedproperly, FocalPoint allows you to replicateyour very best people (Khan, 2002). A second example and, again, perhaps themost ambitious of all, concerns DARPA, who are in the process of developing anUnmanned Combat Air Vehicle (UCAV).According to Boeing (2002), the UCAVsystem is designed to ‘prove the technical

feasibility of multiple UCAVs autonomously

performing extremely dangerous and high-

priority combat missions.’ In a typicalmission scenario, ‘multiple UCAVs will be

equipped with pre-programmed objectives

and preliminary targeting information from

ground-based mission planners. Operations

can then be carried out autonomously, but

can also be revised en route by UCAV

controllers should new objectives dictate.’

If the program is a success, the US DoDexpects to begin fielding UCAV weaponsystems in the 2008 time-frame.

3.3.5.1 Robot teams

Expanding upon the concept of collaborationhighlighted above, one area of AI that isshowing much promise is ‘ubiquitouscomputing’ using information artefacts:future forms of everyday objects thatrepresent a merging of current everydayobjects with the capabilities of informationprocessing and exchange. For example, theEU-funded initiative of the InformationSociety Technologies (IST) researchprogramme aims to show how such artefactscan be made to work together, and inparticular how they provide behaviour or


functionality that exceeds the sum of theirparts (The Disappearing Computer, 2003). Itis from these ideas that the concept of robotteams begins to emerge. Robot teamspotentially have applications in a wide rangeof areas. This is because robots working inteams ‘allow for solutions in which

knowledge, expertise, and motor capability

may be distributed in time and space’ (Maes,1994). Thus, while individual robots mayonly have limited capacity, robots workingtogether in groups might be able to performcomplex tasks. These include militarysurveillance, mine removal, automatedhousehold tasks, large scale laboratoryprojects (such as those used in the HumanGenome Project) and assembly. In this way,most military planners believe that robotsand remote-controlled sensors represent thefuture of information collection on thebattle-field (Jeremiah, 1995).

3 . 3 . 6 Corporate funding

While Section 3.3.5 has demonstrated thesignificant commercial interest in AI, thepicture for corporate investment in this areais a far less coherent. To date, unlike the fieldof nanotechnology, no significant overview ofAI funding seems to exist in the literature.Having said this, however, the level ofcorporate support for AI applicationdevelopment is, in all likelihood,considerable: according to Henry McDonald,Director of the NASA Ames ResearchCentre, one-third of computer-sciencefunding comes from government and two-thirds from industry (cited in Krill, 2002).This is not to say, though, that the interestsof the scientific and business worldsnecessarily concur; while AI may pose manyfascinating questions for the former, suchtechnology has to be commercially viable inthe latter (Broersma, 2001). For this reason,no industry has yet identified a strong motivefor developing strong AI and it is unlikelythat scientists and business people will getany closer together in the future. The centralfocus here, then, must be on the utility of

products, rather than their degree ofintelligence.

As alluded to above, one of the mostc o m m e rcially valuable frontiers of AI is e - c o m m e rce, where technicians are hopingto make the online world simpler and morecapable at the same time. Robots, too, arepotentially big business for the hi-techcompanies pre p a red to invest in them:investment in robots world-wide incre a s e dmarkedly during 2000, with almost 100,000new units being installed, raising the totalstock of robots to 750,000 at the end of2000 (The Economist, 2001).

3.3.6.1 The US and Japan

In general, the US is more widely re g a rded forits private software development than it is forits hard w a re, for which Japan is most highlythought of (Shim, 2002). Indeed, as notede a r l i e r, the number of US AI-related patents in existence increased from 100 in 1989, to1,700 in 1999. Private firms, including larg em a n u f a c t u rers of electronics and computers, as well as major users of IT, hold a vastmajority of these patents. The top three ofthese are IBM (297 patents), Hitachi (192) and Motorola (114), although another 17companies make an appearance on the list7.S i m i l a r l y, many of Japan’s major companieshave plans for AI. According to Shim (2002),the trick for major companies ‘is to time things

right so as to be on the cutting edge of the next

big thing.’ For example, one area in whichSony – one of the most successful companies in the history of consumer electronics – hasinvested in heavily is the home-robot market,t h rough its Entertainment Robot Americadivision. Sony’s latest development in this are ac o n c e rns Aibo Recognition, a mechanical doggranted with the ability to recognise its owner’sname, voice and face, as well as automaticallyre c h a rge itself. By infusing Aibo with incre a s e dAI, such as voice and face recognition, thehope is to give Aibo owners the ability tointeract with a robot at an unprecedented level (Spooner, 2002).

54

3.4 Reality and Hype

3.4.1 Introduction

The kinds of applications outlined abovenecessarily rely to some degree on weak AI.It might seem paradoxical, then, when oneconsiders that it is the area of strong AI thatfeatures more prominently in the publicimagination. To begin with, it is a wellknown fact that many revered members ofthe academic community deem theachievement of machine intelligence reaching,or even surpassing our own, as aninevitability (Barry, 2001). Most famously,this category includes Ray Kurzweil, inventorof the first reading machine for the blind,who believes that ‘within 30 years, we will

have an understanding of how the human

brain works that will give us templates of

intelligence for developing strong AI’ (citedin Anderson, 2001). In fact, as this sectionmakes clear, the future of strong AI is highlyuncertain, with considerable controversypresent within the literature concerningwhether it is even possible or not. Theprimary aim here, then, is to consider thetechnological and philosophical constraintswithin the field. From this, it should be clearthat the issues raised by the possibility ofstrong AI are so fundamental that they crossmany academic boundaries, includingphilosophy, sociology and psychology.

3 . 4 . 2 Barriers to strong AI

The standard test against which the possibilityof strong AI is often judged concerns AlanTu r i n g ’s 1950 article, Computing Machinery

and Intelligence, in which the author discussesthe conditions for considering a machine to beintelligent (Turing, 1950). He argues that if amachine could successfully pretend to behuman to a knowledgeable observer then youc e rtainly should consider it intelligent( M c C a rt h y, 2003). This test would satisfy mostpeople but not all philosophers, some of whichhave challenged the ‘inevitable’ achievement ofs t rong AI based upon the assertion that thehypothesis of strong AI is itself false.

One famous sceptic of AI is Hubert Dreyfus,who says that a computer will never beintelligent unless it can display a goodcommand of common-sense (Dreyfus, 1992).Dreyfus then follows up by saying thatcomputers will never be able to fully graspcommon-sense, since much of our common-sense is on a ‘know-how’ basis. For example,the notion that one solid cannot easilypenetrate another is common-sense, yet theknowledge required to ride a bicycle is notsomething you can gain from a book, orfrom someone telling you. You can onlylearn through experience. Thus, since currentcomputers can only really ‘represent’ things,the possibility of taking a skill, emotion, orsomething else equally abstract, andchanging it into a series of zeros and ones is,according to Dreyfus, close to impossible(Matthews, 1999). A second famous doubteris John Searle, who, with his Chinese Roomanalogy, has responded directly to Turing(cited in Goodwins, 2001):

‘ Take a room with two slots in the wall, an

English-speaking man inside and a ru l e b o o k .

The rulebook tells him how to deal with

Chinese sentences that are pushed through the

slot – how to choose characters with which to

re p l y, and what order to send them back out

t h rough the second slot. The responses may be

p e rfect Chinese, but it does not logically follow

on that the man is actually understanding the

language as a native speaker would, rather

than merely processing it.’

Although a number of convincing rebuttalsto the kinds of philosophical argumentspresented above exist, there can be no doubtthat such positions present intellectuallypowerful barriers to the ultimate goal of AIresearch. Following on from this, it mightappear that opinion in this area is neatlypolarised. However, the picture issignificantly complicated by the fact thatmany researchers consider strong AI asneither particularly likely nor even desirable.In fact, many of the present obstacles to


strong AI research are far more mundane,having been developed as a result of newscientific interest in the mechanisms of thebrain and the way they learn, evolve anddevelop intelligence from a sense of beingconscious. To begin with, althoughcomputers are certainly becoming faster, suchachievements do not necessarily correspondwith computers becoming more intelligent.For, as described by Jaron Lanier (cited inHo, 2002c) as the ‘great shame’ of computerscience, Moore’s law in hardwaredevelopment must be starkly contrasted withthe fact that computer engineers do not seemto be able to write software much better ascomputers get more advanced.

So far, this report has largely focused onways in which scientists model part of whatwe know about our capabilities as sentientbeings, rather than attempting to providetrue sentience. However, even if the ability toprogramme software advances rapidly withinthe next few decades, it seems likely that theAI laboratories of the day will be incapableof providing the kind of environmentnecessary for generating anything resemblingwell-rounded intelligence. This idea stemslargely from the work of Rodney Brooks ofthe MIT who has worked hard in recentyears to challenge prevailing attitudestowards AI research8. Humphrys (1997)builds on these ideas by asserting that youcan’t expect to build a single, isolated AIalone in a laboratory and expect to simulatemuch intelligence. This is because, unless AIsare provided with space in which to evolve arich culture, with repeated social interactionwith things that are like them, you cannotreally expect to get beyond a certain stage.

In addition to software development,significant challenges also exist in thedevelopment of more artificially intelligentrobots. For example, while computer visionis good at certain tasks, there also are manythings it is not particularly good at, such asgeneral object recognition. According to

Brooks (2002), computer vision systems cando a few things with great skill, but still after40 years of effort they are not good at thethings humans and many animals doeffortlessly. Secondly, robots lack thedexterity of the human hand, a primaryingredient in the types of manufacturing thathave moved to low-cost locations. Accordingto Brooks, ‘low-cost dextrous manipulation’

is essential if progress is to be made. Atpresent, however, even high-cost dextrousmanipulation is beyond researchers.Furthermore, such challenges are unlikely tobe met in the next few years, possiblyrequiring 30–40 years before suchtechnologies are refined.

3 . 4 . 3 A future for strong AI?

In spite of the many fundamental barr i e r shighlighted above, the fields of AI androbotics are replete with many wonderf u l l yinventive predictions, a domain where re a l i t yand science fiction often meet. Indeed, it islikely that in the next two decades ‘we’ll see

m o re and better capabilities that we tend to

attribute as awareness’ ( H e n d l e r, 2000).H o w e v e r, it is unlikely that machines will everhave human awareness in the philosophicalsense of the term, although they may comeclose in the long term. Rather, we can expectto see classical AI going on to produce moreand more sophisticated applications inrestricted domains, such as expert systems,chess programs and Internet agents. At thesame time, the next 30 years will pro d u c enew types of animal-inspired machines thata re more ‘messy’ and unpredictable than anywe have seen before – less rationallyintelligent but more rounded and whole( H u m p h rys, 1997).

One potentially far-reaching developmentinvolves side-stepping the seemingly polarisedweak/strong AI debate through thedevelopment of cyborg technology, theapplications of which could lead to humanshaving certain physiological processes aidedor controlled by mechanical or electronic

56

devices. The most high-profile demonstrationin this area concerns ‘robo-rat’, which,through the implantation of electrodes intothe parts of the brain responsible for sensingreward and for stimulation from the left andright whiskers, has been successfully guidedby a human controller (Graham-Rowe,2002). A similar experiment has also beendemonstrated by Steve Potter, Professor ofBiomedical Engineering at the GeorgiaInstitute of Technology, who has developed a‘rat-controlled robot’ (Cameron, 2002). Thisdevice results from placing a droplet ofsolution containing thousands of rat neuroncells onto a silicon chip and then relaying theresulting electrical activity to a robot. Therobot then manifests these signals withphysical motion, each of its movements adirect result of neurons communicating withneurons. Such examples of merging computerchips with living tissue may seem crude, butare described by scientists as ‘momentous’ –an event comparable to the first organtransplant or cloned animal (Philipson,2001). This is because such experiments openup the possibility of using computertechnology to supplement humanintelligence, rather than replace it.

In conclusion, then, we will not see full AI inour lives. The reason is that there is noobvious way of getting from here to there –f rom the rather useless robots and brittles o f t w a re programs in existence nowadays tohuman-level intelligence. A long series ofconceptual bre a k t h roughs are needed, and thiskind of thinking is very difficult to timetable.

3.5 Concerns

3 . 5 . 1 I n t r o d u c t i o n

The fields of strong AI and robotics aregenerally regarded as controversial becauseof their far-reaching social, ethical, andphilosophical implications. Researchmanagers are in no doubt that suchcontroversy has affected the fundingenvironment for AI and the objectives of

many research programmes (NRC, 1999).However, in general, less attention is paid tothe implications of weak AI, even thoughmany of the applications of this field, asdemonstrated above, are in operation today.In other words, it should be recognised thatmany of the concerns described below do notrely on the long-term development of strongAI as popularly imagined. As for Section 2.5on nanotechnology then, this section, as wellas considering the connotations of AI, willattempt to distinguish between short- andlong-term concerns that advancements in thisarea will surely bring.

3 . 5 . 2 Predictive intelligence

A c c o rding to Kirsner (2002), the technologyw o r l d ’s big debate for 2003 will centre onp redictive intelligence. This aspect of AI,a l ready touched upon above, concerns theability to use software running on powerf u lcomputers to analyse information about onesprior behaviour. In the private sector,companies are already using pre d i c t i v eintelligence to analyse data profiles and solvem o re mundane business problems. Theseinclude Epsilon – a database marketingcompany based in the US, which have beencombing through transactional data since the1980s to help its customers market moree ffectively – along with other projects designedto identify which customers are more likely tospend the most money (Kirsner, 2002).

The most dramatic example of this is pro v i d e dby the US DoD, which has established are s e a rch group to develop technology fori n f o rmation gathering and analysis on a hugescale. Its goal is to mine data sources all overthe world – including government andc o m m e rcial stores of personal information –to look for terrorists and terrorist thre a t s(Anthes, 2002). This programme includes therecently-established controversial To t a lI n f o rmation Aw a reness (TIA) office whichaims to ‘revolutionise the ability of the US todetect, classify and identify foreign terro r i s t s ,decipher their plans, and take timely action to


p re-empt and defeat terrorist acts.’ The toolswhich the TIA intends to develop to achievethis rely to a large extent on new AItechnologies. These include ‘entity extractionf rom natural language text’ and ‘biologicallyi n s p i red algorithms for agent contro l . ’F u rt h e rm o re, one of the TIA’s 13 subdivisions,the Human Identification at a Distance(HumanID) programme, is releasing contractsfor face, iris and gait recognition. Another ofthe subdivisions, FutureMap, will concentrateon market-based techniques for avoidingsurprise and predicting future events( H e rt z b e rg, 2002).

A second programme, called EvidenceExtraction and Link Discovery (EELD), aims to develop technology for ‘automated

discovery, extraction and linking of sparse

evidence contained in large amounts of

classified and unclassified data sources’

(Anthes, 2002). In order to achieve this,EELD will have to develop detectioncapabilities to extract relevant data andrelationships about people, organisations and activities from huge volumes of data.

A p a rt from the sheer ambitiousness of thep rogrammes, TIA and EELD have generatedc o n c e rn mainly in relation to theirimplications for infringing individual andg roup privacy, and the possibility of suchi n f o rmation being handled carelessly or evenleading to malevolence. Indeed, it only takes amoment of reflection to consider that nearlye v e ryone in modern society has at least onefact about themselves to hide. And yet, inspite of these well-founded concerns, both theTIA and EELD are already in activedevelopment; in response, Hert z b e rg (2002)recommends that, at a minimum, a temporaryshutdown of the EELD system pending somes o rt of congressional review and the cre a t i o nof safeguards is highly desirable.

3.5.3 AI and robotic autonomy

Many of the major ethical issues surro u n d i n gA I - related development hinge upon the

potential for software and robot autonomy. In the short term, some commentatorsquestion whether people will really want tocede control over our affairs to an art i f i c i a l l yintelligent piece of software, which might evenhave its own legal powers. Broersma (2001)believes that, while some autonomy isbeneficial, absolute autonomy is frightening.For one thing, it is clear that legal systems arenot yet pre p a red for high autonomy systems,even in scenarios that are relatively simple to envisage, such as the possession of personali n f o rmation. In the longer- t e rm, however, inwhich it is possible to envisage extre m e l yadvanced applications of hard AI, seriousquestions arise concerning military conflict,and robot ‘take-overs’ and machine rights.Each of these is dealt with in turn below.

3.5.3.1 AI and military conflict

This re p o rt shows that the military interest in AI is significant. However, as pointed outabove, the difficulties involved in achievinganything resembling hard AI surely mean thatany such system will be subject to re l i a b i l i t yc o n c e rns. This idea is not new; the issue ispicked up by Thompson as early as 1977, whosets out his concerns re g a rding existing andplanned uses of computer technology as part of nuclear weapons systems. More generally, it is his belief that no computer system has thecapacity to reliably make decisions of there q u i red kind and in the re q u i re dc i rcumstances, nor can one ever be constru c t e d .This is because the complexity and sensitivityof such systems makes exhaustivecharacterisation extremely difficult, and anyresulting mistakes cannot be corrected via theusual process of use, failure and modification.M o re re c e n t l y, the controversial US NationalMissile Defence programme, which is beingdesigned using the latest AI technology,p rovides a second example. The system issupposed to dispense ‘kill power’ based on an ability to recognise incoming missiles in amatter of seconds and then decide whether tod e s t ro y, intercept of ignore them (Newquist,1987). However, serious concerns are alre a d y

58

being voiced based upon the workability ofsuch a system. This is because, while testingmay be possible for an autonomous tank andother weapons of the electronic battlefield, it isnot feasible for National Missile Defence. Sucha system can only be realistically evaluated inactual combat (Augarten, 1986). Moref u n d a m e n t a l l y, significant moral diff i c u l t i e sarise out of human distaste for autonomousweapons. Gary Chapman (2000) summarisesthis concern well:

‘[Such arms] are a revolution in warf a re in that

they will be the first machines given the

responsibility for killing human beings without

human direction or supervision. To make this

m o re accurate, these weapons will be the first

killing machines that are actually pre d a t o ry,

that are designed to hunt human beings and

d e s t roy them.’

Indeed, the UCAV example provided abovedemonstrates that potentially, in battle,humans may be taken out of the decision-making loop and still be on the receiving end –w h e re the ‘kill power’ goes.

3.5.3.2 Robot ‘take-over’ and machine rights

Such issues of predatory machines are boundto raise concern over the scenario of AIsovertaking humankind and thus somehowcompeting with him. This idea has oftenbeen popularised by classic science fictionworks and populist academics, such asProfessor Kevin Warwick, Professor ofCybernetics at the University of Reading,UK, who has repeated this beliefs concerningrobot ‘take-over’ on many occasions in thepress, in his books, and on television andradio. Consider the following letter fromNicholas Albery (1999) of the Institute ofSocial Inventions. Published in New Scientistand entitled Robot Terror, Albery seekssupport for the following petition:

‘In view of the likelihood that early in the

next millennium computers and robots will

be developed with a capacity and complexity

greater than that of the human brain, and

with the potential to act malevolently

towards humans, we, the undersigned, call

on politicians and scientific associations to

establish an international commission to

monitor and control the development of

artificial intelligence systems.’

It is this kind of claim that seems to infuriatemany in the AI scientific community. ChrisMalcolm (2001) of the School of ArtificialIntelligence at Edinburgh University, forexample, describes belief in the robot take-over scenario as ‘dangerous’ and‘misleading’. He points out that publicoverreaction to AI stems from an assumptionthat something which displays some of theattributes of creaturehood must possess allthe attributes of creaturehood. In his words:

‘Intelligence is no more enough to make a

real creature than is fur and beady eyes. No

matter how much intelligence is added to

your word processor it is not going to sulk

and refuse to edit any more letters if you

don’t improve your spelling...Our problem is

that while we have got used to the idea that

teddy bears are not real even though we may

be in the habit of talking to them at length,

we are not used to contraptions being

intelligent enough to talk back, and are

willing to credit them with possession of the

full orchestra of creaturehood on hearing a

few flute-like notes.’

Perhaps the most measured assessment of thepossibility of tyrannical take-over to datestems from the work of Whitby and Oliver(2001), who, in addition, to the classic worstcase scenario, focus on the more subtle ideasof ‘cultural reliance’ and ‘co-evolution’. Withregard to the former, the authors concludethat: ‘although not obviously misguided or

incoherent, predictions of tyrannical take-

over are wrong. This is due to a number of

possible failsafe methods, such as buddy

systems, ethical systems programming, and

perhaps most importantly, humans as final


arbitrators in decision making.’ In any case,it is not clear in the first place whyintelligence should necessarily be regarded as synonymous with aggression. On the otherhand, cultural reliance, in which humanssomehow allow a position of dependency onAI and robotics to develop, and co-evolution,in which human and machine becomeinextricably intertwined, are regarded asmore probable.

The strong public reaction to machine take-over appears, then, not to be well founded.However, if it is possible to agree, forargument's sake, that humankind will be able to create a truly intelligent machine, amuch deeper issue arises: how will a sentientartificial being be received by humankindand by society? Barry (2001) asks pertinentquestions: ‘Would it be forced to exist like its

automaton predecessors who have effectively

been our slaves, or would it enjoy the same

rights as the humans who created it, simply

because of its intellect?’ This is an enormousquestion that touches religion, politics andlaw, but to date little serious discussion hasbeen given to the possibility of a newintelligent species and to the rights anautonomous sentient might claim.

3.6 DiscussionThe short-term concerns surrounding AI and robotics are mainly ethical in nature.This is in contrast to nanotechnology, thepotential dangers of which cover a muchlarger spectrum and one that includesenvironmental risk. As shown above, weakAI tends to create concern with respect to its role as a tool for human interaction,throwing up issues of responsibility,privacy and trust. Applications in this areaare emerging all the time, making 2003 theright time to begin public debate over theseconcerns. This is important for three mainreasons. First, there might be a tendency forAI technology to creep into out lives largelyunnoticed. This is because of the well-documented AI effect, due to which the

major applications of AI research are mostlyhidden from view because they are embeddedin larger software systems. Second, many ofthese applications are morally ambiguous – a grey area of ethics that stands in starkcontrast to Isaac Asimov’s famously clear-cutthree laws of robotics9. Third, presumingthat a public debate over AI can be initiated,there is little evidence to date that thisdiscussion will affect military andcommercial interests. Having said that, thereis evidence of some attempt to flesh out acode of professionalism for AI. For example,in reference to AI and responsibility, Whitby(1984) writes:

‘Where an AI system is introduced into any

human system it shall be the responsibility

of the AI professional to ensure that a

human or group of humans within the

system shall take moral and/or legal

responsibility for the human consequences

of any malfunction of the AI system.’

However, there is little sign in the literaturethat suggests these ideas have been followedup on.

Strong AI, on the other hand, asks muchmore fundamental questions as the fieldnecessarily deals with human/machinerelationships per se. As a consequence, thekinds of tools that might be necessary tobegin debate over strong AI are not even hereyet, so great are the implications. However, itis likely that this technology will not occur inour lifetimes; regardless of how oftenProfessor Warwick is presented as an AIexpert, the fact remains that his opinions arefar removed from the majority view of the AI community (Colton, 2001). On the otherhand, this report is by no means intended todownplay such potentially revolutionarydevelopments as ‘mere’ science fiction. For,if the long-term potential of AI was to berealised, then it would surely have ademonstrable impact in a whole range ofindustrial and, in particular, service sectors.

4. Conclusion

60

This report began by stressing the need toprovide background information onnanotechnology and AI. In doing so, it washoped that the prospects of these emergingtechnologies to affect quality of life in thecoming decades could be realisticallyassessed. One consequence of providing such an overview is that there can be nodecisive conclusions as such; the industriescharacterised here are too dynamic anduncertain to generate any real sense ofresolution. However, it is possible tohighlight a number of important differencesand similarities between nanotechnology and AI which go some way to shedding more light on their character.

Perhaps the greatest contrast between thetwo industries concerns public interest.Indeed, as this report has demonstrated,nanotechnology is widely regarded as a ‘new’ and exciting branch of science andtechnology. This belief has contributed to the massive period of growth that this high-profile and wide-ranging field is currentlyenjoying. AI, on the other hand, is viewed bymany as an highly specialised and unprovendiscipline. One reason for this concerns thegross over-optimism that characterised theindustry in the 1960s and 1980s. Anotherreason reflects the AI community’s seeminglyinsurmountable difficulty in publicising itsown achievements without whipping upgeneral anxiety over machine superiority.The upshot of all this has been the field’sstruggle to attract funding in the past and it is likely that this trend will continue forsometime into the foreseeable future.

Revealing similarities also exist betweennanotechnology and AI. There has beenmuch talk recently regarding the convergenceof traditionally separate scientific fields, inparticular the blurring of the boundariesbetween the physical sciences and lifesciences – perhaps even the first step towardsthe long sought after unification of physics,chemistry and biology (Howard, 2002). For

example, the concourse of nanoscience,biotechnology, IT, and cognitive science(‘NBIC’) was discussed during a December2001 NSF workshop. NBIC, it was agreed‘could achieve a tremendous improvement in

human abilities, societal outcomes, the

nation’s productivity and the quality of life’

(Roco and Bainbridge, 2003). In some ways,the above conclusion is hardly surprisinggiven the ambitious and broad scope of thetechnologies discussed in this report. Aspointed out above, ‘convergence’ largelyarises from the wide availability oftechniques and tools on offer today – the realinnovation stems from the process ofbringing individuals from traditionallyseparate disciplines together.

Most importantly for convergence here, it is possible that developments innanotechnology could lead to advances in AI through improvements in computerminiaturisation, performance, or architecture(but see Section 3.4.2 on barriers to strongAI), or through the sensor interface. Inaddition, it seems fair to assume that anyfuturistic nanobots would have to be imbuedwith a reasonable degree of AI. A second,more contentious similarity concernsreinvention. As demonstrated in this report,the ‘rediscovery’ of AI has been a virtualnecessity for the survival of the industry; for nanotechnology the phenomena is lessobvious but is arguably there all the same.That is, as a natural extension of themicromechanical and MEMS research thatbegun in the 1960s, nanoscience is hardly‘new’ as such; rather, ‘nano’ can be viewed as a useful tag with which to boost funding.Just what the consequences of this strategywill be, it is hard to tell. Ironically, AIprovides an excellent example of a promisingscientific discipline that has often resulted indisappointment. Whether the same happensto nanotechnology remains to be seen.

The second consequence of providing anoverview is that certain elements of


nanotechnology and AI development arebound to be overlooked. First, the difficultiesof drawing out accurate statistics forcorporate R&D have already been alluded to earlier. Second, there are wide rangingapplications across the economy for sensorsthat can support industrial processes and beincorporated into new or existing products(Miles and Jarvis, 2001). The application ofnanotechnology to this area should allow forimprovements in functionality and muchdecreased size. Third, a more in-depthanalysis of environmental concerns iswarranted. This is because publicacceptability of such risk is likely to varyconsiderably in relation to the applicationbeing considered. For example, theapplication of nanotechnology tocomputerisation is less likely to causeconcern than those practices which mightlead to the release of nanoparticles into theenvironment, such as the disposal of nano-based composites. Fourth, it is possible toconceive of a number of environmentalgoods that may arise. For example, thepotential for gains in energy generation andefficiency have already pointed out above(Section 2.3.4.), and it is conceivable thatdramatic improvements in environmentalsensing and modelling could also beachieved. However, any pervasive diffusionof nano- and AI-based technologies in thecoming decades is bound to have asignificant effect on the demand for resources

by industry, transport and the domesticsector. The way in which these morefundamental changes might impact on theenvironment would have to form the basis of a much larger technology assessment, inwhich long-term structural changes to globalindustry and commerce were considered.

F i n a l l y, it is easy to overlook the lessons thatattitudes towards technological developmentteach us about human nature. This re p o rt hasl a rgely relied upon the technique of lookingahead, identifying technological possibilities,and assessing the likelihood of successfullymoving towards their realisation. Significantly,this process mirrors that of technologicalinnovators, a kind of thinking that oftentranslates into the belief that technologicaldevelopment is autonomous – the ultimateself-fulfilling pro p h e c y. To some extent we area l ready on this road. Most technologiesc o v e red in this re p o rt are within the bounds ofc u rrent scientific possibility and it is just amatter of time, eff o rt and expenditure beforethey are realised. However, the contrastingf o rtunes of the nanotechnology and AIindustries remind us that much of thisp ro g ress hinges on public appro v a l .U l t i m a t e l y, a 21st-Century acceptance modelcalls for technological innovations to bereceived on a voluntary basis where thep e rceptible usefulness of new technologyp roducts are balanced against associated risksthat are shown to be manageable.

E n d n o t e s

62

1 – 9

1 Grove-White, R., Macnaghton, P. and Wynne B. (2000). Wising Up: The Publicand New Technologies, Lancaster, UK: IEPPP, Lancaster University.

2It is worth bearing in mind when consulting this type of information that, given

the difficulty of even agreeing on what constitutes nanotechnology, many of thenumbers presented below should be treated with caution (Roman, 2002).

3 For a detailed breakdown, see the NNI’s own report at:http://www.nano.gov/2003budget.html

4 For a detailed description of the history of AI, see the University of Edinburgh’sDivision of Informatics Website at:http://www.dai.ed.ac.uk/AI_at_Edinburgh_perspective.html

5 For more information, see Ruthfield, 1995.

6 See Honda’s Website for more information at: http://world.honda.com/robot/

7 For details, see National Research Council, 1999.

8 For a full description of this paradigm shift, see Humphrys, 1997.

9 The Three Laws of Robotics are:

a. A robot may not injure a human being, or, through inaction, allow ahuman being to come to harm;

b. A robot must obey the orders given to it by human beings except wheresuch orders would conflict with the First Law;

c. A robot must protect its own existence as long as such protection does notconflict with the First or Second Law.

For a fuller explanation see:http://whatis.techtarget.com/definition/0,,sid9_gci520366,00.html


Albery, N. (1999). Robot Terror.New Scientist; 161 (2171): 49.[letter]

Aleksander, I. (2002). The NextBig Thing. BBCi & The Open

University [online].http://www.open2.net/nextbigthing/ai/hear_the_arguments/argument3.htm

Anderson, K. (2001). PredictingAI’s Future. BBC News; 21 Sep2001 [online].http://news.bbc.co.uk/1/hi/in_depth/sci_tech/2001/artificial_intelligence/1555742.stm

Antón, P.S. Silberglitt, R. andSchneider, J. (2001). The Global

Technology Revolution:

Bio/Nano/Materials Trends and

Their Synergies with Information

Technology by 2015. SantaMonica, CA: RAND. DocumentNumber: MR-1307-NIC.http://www.rand.org/publications/MR/MR1307/

Anthes, G.H. (2002). GlobalSurveillance: The Government’sPlan. Computerworld; Security:Privacy, 25 Nov 2002 [online].http://www.computerworld.com/securitytopics/security/privacy/story/0,10801,76117,00.html

Augarten, S. (1986). CanComputers be Failsafe? PCMagazine; 5 (1): 97.

Barry, R. (2001). Sentience: TheNext Moral Dilemma. ZDNet

UK; News: 24 Jan 2001 [online].http://news.zdnet.co.uk/story/0,,s2083942,00.html

.BBC (2002). Computers ReachOne Million Mark. BBC News; 1Jul 2002 [online].http://news.bbc.co.uk/1/hi/sci/tech/2077986.stm

Boeing (2002). UnmannedCombat Air Vehicle (X-45).Boeing [online].http://www.boeing.com/phantom/ucav.html

Broersma, M. (2001). AI GetsDown to Business. ZDNet UK;News: 23 Jan 2001 [online].http://news.zdnet.co.uk/story/0,,s2083916,00.html

Brooks, R. (2001). Interviewed in:Anderson, K. Predicting AI’sfuture. BBC News; 21 Sep 2001[online].http://news.bbc.co.uk/1/hi/in_depth/sci_tech/2001/artificial_intelligence/1555742.stm

Brooks, R. (2002). Interviewed in:Jusko, J. The Robot Evolution.Industry Week; 1 Nov 2002[online].http://www.industryweek.com/CurrentArticles/Asp/articles.asp?ArticleId=1356

Buchanan, B. and Uthurusamy, S.(1999). The InnovativeApplications of ArtificialIntelligence Conference. AI

Magazine; 20 (1): 11–12.

Cameron, D. (2002). Rat-BrainedRobot. Technology Review; 18Dec 2002 [online].http://www.technologyreview.com/articles/print_version/wo_cameron121802.asp

Chapman, G. (2000). Quoted in:Matthews, J. (2000b). MilitaryApplications of AI. Generation5

Essays [online].http://www.generation5.org/app_military.shtml

Chaudhari, P. (2001). FutureImplications of Nanoscale Scienceand Technology: Wired Human,Quantum Legos, and an Ocean ofInformation. In: M.C. Roco, W.S.Bainbridge (editors). Societal


Nanotechnology. Arlington, VA,USA: National ScienceFoundation; pp75–9.http://www.wtec.org/loyola/nano/NSET.Societal.Implications/nanosi.pdf

Colton, S. (2001). Cyborg Off HisChristmas Tree. Times Higher

Education Supplement; 21 Dec2001: p14.http://www.aisb.org.uk/articles/computology.html

Colvin, V. (2002). ResponsibleNanotechnology: Looking Beyondthe Good News. EurekAlert!;Nanotechnology In Context: Nov2002 [online].http://www.eurekalert.org/context.php?context=nano&show=essays&essaydate=1102

Compano, R. editor (2001).Technology Roadmap for

Nanoelectronics. Luxembourg:Office for Official Publications ofthe European Communities.

A – CR e f e r e n c e s

64

Compano, R. and Hullman, A.(2002). Forecasting theDevelopment of Nanotechnologywith the Help of Science andTechnology Indicators.Nanotechnology; 13 (3): 243–7.

Copeland, B.J. (2000). What isArtificial Intelligence?AlanTuring.net; ReferenceArticles: May 2000 [online].http://www.alanturing.net/turing_archive/pages/Reference%20Articles/what_is_AI/What%20is%20AI02.html

Cowen, R. (2002). Ribbon to theStars. Science News; 162 (14):218. http://www.sciencenews.org/20021005/bob9.asp

Dalesio, E.P. (2002). ‘Text mining’Software Business Grows. Sun-

Sentinel; 4 Mar 2002 [online].http://www.sun-sentinel.com/business/local/sfl-0304textmining.story?coll=sfla-business-headlines

Department of Defense (2002).Nanotechnology Research AwardsAnnounced. DefenseLINK; News:23 Feb 2001 [online].http://www.defenselink.mil/news/Feb2001/b02232001_bt079-01.html

Department of Trade and Industry(2002). New Dimensions for

Manufacturing: UK Strategy for

Nanotechnology. Report of theUK Advisory Group onNanotechnology Applications;June 2002. UK: Department ofTrade and Industry.http://www.dti.gov.uk/innovation/nanotechnologyreport.pdf

Dertouzos, M.L. (1999). TheFuture of Computing. Scientific

American; 281 (2): 52–5.http://www.sciam.com/issue.cfm?issuedate=Aug-99

Disappearing Computer, The(2003). Mission Statement. The

Disappearing Computer [online].http://www.disappearing-computer.net/

Doering, R. (2001). SocietalImplications of Scaling toNanoelectronics. In: M.C. Roco,W.S.Bainbridge (editors). Societal



Doyle, J. and Dean, T. (1996).Strategic Directions in ArtificialIntelligence. ACM Computing

Surveys; 28 (4): 653–70.http://medg.lcs.mit.edu/ftp/doyle/sdai96.html

Dreyfus, H. (1992). What

Computers Still Can’t Do: A

Critique of Artificial Reason .Cambridge, MA, USA:Massachusetts Institute ofTechnology.

Economist, The (2001). Robots.The Economist.com; 29 Nov 2001[online].http://www.economist.com/printedition/displayStory.cfm?Story_ID=887290

Economist, The (2002). Trouble inNanoland. The Economist.com; 5Dec 2002 [online].h t t p : / / w w w. e c o n o m i s t . c o m / s c i e n c e /d i s p l a y s t o ry. c f m ? s t o ry _ i d = 1 4 7 7 4 4 5

Engineering and Physical SciencesResearch Council (2003). EPSRC

Web Site [online].http://www.epsrc.ac.uk

ETC Group (2002a). No SmallMatter! Nanotech ParticlesPenetrate Living Cells andAccumulate in Animal Organs.ETC Group Communique ; (76)May/June 2002 [online].http://www.etcgroup.org/documents/Comm_NanoMat_July02.pdf

ETC Group (2002b). NanotechTakes a Giant Step Down! ETC

Group News Release; 6 Mar 2002[online].http://www.etcgroup.org/documents/nr2002march6.pdf

ETC Group (2002c). PatentingElements of Nature. ETC Group

Geno-type; 25 Mar 2002 [online].http://www.etcgroup.org/documents/nanopatentsgeno.rtf.pdf

European Environment Agency(2003). EEA MultilingualEnvironmental Glossary.European Environment Agency

[online].http://glossary.eea.eu.int/EEAGlossary/P/precautionary_approach

Fishbine, G. (2002). Investor’s

Guide to Nanotech and

Micromachines. Chichester, UK:John Wiley.

C – F


FOLDOC (2003). Free On-Line

Dictionary of Computing

[Online].http://wombat.doc.ic.ac.uk/foldoc/

Forrest, D. (1989). RegulatingNanotechnology Development.Foresight Institute; 23 Mar 1989[online].http://www.foresight.org/NanoRev/Forrest1989.html

Freitas, R.A. (2000). Some Limitsto Global Ecophagy by BiovorousNanoreplicators, with PublicPolicy Recommendations.Foresight Institute; Mar 2000[online].http://www.foresight.org/NanoRev/Ecophagy.html

Fried, J. (2002). Japan SeesNanotech as Key to Rebuilding itsEconomy. Small Times; 7 Jan2002 [online].http://www.smalltimes.com/document_display.cfm?document_id=2843

Glinos, K. (1999).Nanotechnology: Blurring theBoundaries at the Atomic Scale.RTD info; (21) Feb 1999 [Online].http://www.europa.eu.int/comm/research/rtdinf21/en/dossier1.html

Goodwins, R. (2001). TheMachine that Wanted to be aMind. ZDNet UK; News: 23 Jan2001 [online].http://news.zdnet.co.uk/story/0,,s2083911,00.html

Gorman, J. (2002). Taming High-Tech Particles: Cautious Steps intothe Nanotech Future. Science

News; 161 (13): 200.http://www.sciencenews.org/20020330/bob8.asp

Graham-Rowe, D. (2002). “Robo-rat” Controlled by BrainElectrodes. New Scientist; 417: 37.http://www.wireheading.com/roborats/ratbot.html

Grosz, B. and Davis, R. editors(1994). A Report to ARPA on

Twenty-First Century Intelligent

Systems. Menlo Park, CA, USA:American Association forArtificial Intelligence.http://www.aaai.org/Resources/Policy/arpa-report.html

Grupp, H. editor (1993).Technologie am Beginn des 21.

Jahrhunderts. Heidelberg: PhysicaVerlag. Cited in: Compano, R. andHullman, A. (2002).

Gsponer, A. (2002). From the Labto the Battlefield? Nanotechnologyand Fourth-Generation NuclearWeapons. Disarmament

Diplomacy; (67) Oct–Nov 2002[online].http://www.acronym.org.uk/dd/dd67/67op1.htm

Harper, T. (2002). TheNanotechnology Arms Race: WhyNobody Wants to be Left Behind.nanotechweb.org; 14 Nov 2002[online].http://nanotechweb.org/articles/column/1/11/1/1

Hay, J.N. and Shaw, S.J. (2000). AReview of Nanocomposites 2000.Proceedings of the

Nanocomposites 2000

Conference; 6–7 Nov 2000,Brussels, Belgium. Farnborough,UK: Defence Evaluation andResearch Agency.http://www.nano.org.uk/nanocomposites_review.pdf

Helsel, S. (2002). Are Corporateand Government FundingOverlooked?NanoelectronicsPlanet; 26 Jul2002. [online].http://www.nanoelectronicsplanet.com/features/article/0,4028,6571_1433691,00.html

Herrera, S. (2002). No smallmatter. Red Herring; 16 Dec 2002[online].http://www.redherring.com/insider/2002/12/10trends-nanotech121602.html

Hertzberg, H. (2002). Too MuchInformation. The New Yorker; 9Dec 2002. [online]http://www.newyorker.com/talk/content/?021209ta_talk_hertzberg

Hendler, J. (2000). A Chat Aboutthe Future of ArtificialIntelligence. CNN; 1 Jan 2000[online].http://www.cnn.com/COMMUNITY/transcripts/1999/12/hendler/

Ho, M.W. (2002a).Nanotechnology: A Hard Pill toSwallow. Science in Society; (16)Autumn 2002 [online].http://www.i-sis.org.uk/nanotechnology.php

F – H

66

Ho, M.W. (2002b). The BraveNew World Quartet. Science in

Society; (16) Autumn 2002[online]. http://www.i-sis.org.uk/bravenewworld.php

Ho, M. (2002c). Will ComputersBecome Super-Human? Science in

Society; (16) Autumn 2002[online]. http://www.i-sis.org.uk/computersvshumans.php

Holister, P. (2002). Nanotech: TheTiny Revolution. CMP Cientifica;July 2002 [online].http://www.cientifica.info/html/docs/NOR_White_Paper.pdf

Howard, S. (2002).Nanotechnology and MassDestruction: The Need for anInner Space Treaty. Disarmament

Diplomacy; (65) Jul–Aug 2002[online].http://www.acronym.org.uk/dd/dd65/65op1.htm

Hsiung, S. (2002) An Introductionto Artificial Intelligence.Generation5 [online].http://www.generation5.org/aiintro.shtml

Humphrys, M. (1997). AI isPossible…But it Won’t Happen:The Future of ArtificialIntelligence. Proceedings of the

Next Generation Symposium; Aug1997, Jesus College, Cambridge,UK. http://www.compapp.dcu.ie/~humphrys/newsci.html

IT Strategic Headquarters (2001).E-Japan Strategy. Policy Initiative,22 Jan 2001 Tokyo, Japan: ITPolicy Office.http://www.kantei.go.jp/foreign/it/network/0122full_e.html

Jeremiah, D.E. (1995).Nanotechnology and GlobalSecurity. Proceedings of the

Fourth Foresight Conference on

Molecular Nanotechnology;9–11 Nov 1995, Palo Alto, CA,USA. [online].http://www.zyvex.com/nanotech/nano4/jeremiahPaper.html

Joseph, S. (2001). AI. J@pan Inc

Magazine; Nov 2001 [online].http://www.japaninc.com/article.php?articleID=514

Khan, J. (2002). It’s Alive! Wired;10 (3) [online].http://www.wired.com/wired/archive/10.03/everywhere.html

Kirsner, S. (2002). Getting SmartAbout Predictive Intelligence.Boston Globe; 30 Dec 2002[online].http://www.boston.com/globe

Krill, P. (2002). CTO Forum: WillArtificial Intelligence Surpass theHuman Variety? InfoWorld; 11Apr 2002 [online].http://www.infoworld.com/articles/hn/xml/02/04/11/020411hnartificial.xml

LaVan, D.A. and Langer, R.(2001). Implications ofNanotechnology in thePharmaceutics and Medical Fields.In: M.C. Roco and W.S.Bainbridge (editors). Societal



Leo, A. (2001). Get Ready forYour Nano Future. Technology

Review; May 2001 [online].http://www.technologyreview.com/articles/wo_leo050401.asp

Lighthill, J. (1973). ArtificialIntelligence: A General Survey. In:Artificial Intelligence: A Paper

Symposium. London, UK: ScienceResearch Council.

Machan, D. (2002). A Few GoodToys. Forbes Magazine; (12) 9 Dec 2002 [online].http://www.forbes.com/forbes/2002/1209/120.html

Maes, P. (1994). Agents ThatReduce Work and InformationOverload. Communications of theACM; 37 (7): 31–40.http://pattie.www.media.mit.edu/people/pattie/CACM-94/CACM-94.p1.html

Malcolm, C. (2001). RobotsWon’t Rule. School of Artificial

Intelligence, Division of

Informatics, Edinburgh

University; 3 Aug 2001 [online].http://www.dai.ed.ac.uk/homes/cam/Robots_Wont_Rule.shtml

Martel, S. (2000). Introduction tothe NanoWalker: A MiniatureAutonomous Robot Capable ofVarious Tasks at the Molecularand Atomic Scales. Bio-

Instrumentation Laboratory;Projects [online].http://bioinstrumentation.mit.edu

H – M


Matthews, J. (1999). PhilosophicalArguments For and Against AI.Generation5; 13 Dec 1999[online].http://www.generation5.org/ai_phil.shtml

Matthews, J. (2000a).Introduction to Neural Networks.Generation5; 20 Mar 2000[online].http://www.generation5.org/nnintro.shtml

Matthews, J. (2000b). MilitaryApplications of AI. Generation5;8 May 2000 [online].http://www.generation5.org/app_military.shtml

McCarthy, J. (2003). What isArtificial Intelligence? Computer

Science Department, Stanford

University; 29 Mar 2003 [online].http://www-formal.stanford.edu/jmc/whatisai/whatisai.html

McCullagh, D. (2002). ReportCalls for Nanotech Laissez-Faire.CNET News.com; 21 Nov 2002[online].http://www.news.com.com/2100-1023-966766.html

Melymuka, K. (2002). GoodMorning, Dave... Computerworld;11 Nov 2002 [online].http://www.computerworld.com/softwaretopics/software/appdev/story/0,10801,75728,00.html

Miles, I. and Jarvis, D. (2001).Nanotechnology – A Scenario for

Success in 2006. Teddington, UK:HMSO. National PhysicalLaboratory Report Number:CBTLM 16.http://libsvr.npl.co.uk/npl_web/pdf/cbtlm16.pdf

Miller, D. (2001). FiveTechnologies You Need to Know.The Industry Standard; 21 May2001 [online].http://www.thestandard.com/article/0,1902,24308,00.html

Miller, P. (2002). Big, Big Plans forVery Small Things. Green Futures;Jul/Aug 2002: 50–1.http://www.paulmiller.org/Nanotechnology.pdf

Tam, A. (2001). Japan’s NanotechPlayers: Thinking Big. J@pan Inc

Magazine; May 2001 [online].http://www.japaninc.net/mag/comp/2001/05/print/may01p_investor_nanotech.html

Mokhoff, N. (2002). SpeechTechnology Looses its KookyLuster. EETimes; 1 Nov 2002[online].http://www.eetimes.com/story/OEG20021031S0046

nanotechweb.org (2002).Nanocomposites Set to Wrap upthe Packaging Market.nanotechweb.org; 23 Aug 2002[online].http://nanotechweb.org/articles/news/1/8/18/1

National Institute of Informatics(2002). National Institute of

Informatics 2002. Tokyo, Japan:National Institute of Informatics.http://www.nii.ac.jp/publications/nii-yoran/yoran2002-e.pdf

National Research Council(1999). Funding a Revolution:

Government Support for

Computing Research. Washington,DC, USA: National AcademyPress.http://www.nap.edu/readingroom/books/far/notice.html

Nelson, M. and Shipbaugh, C.(1995). The Potential of

Nanotechnology for Molecular

Manufacturing. Santa Monica,CA, USA: RAND. DocumentNumber: MR-615-RC.http://www.rand.org/publications/MR/MR615/

Newquist, H. (1987). Star Wars:When Imperfect Man Strives toMake Perfect Machines.Computerworld; 30 Nov 1987:17. http://www-cse.stanford.edu/classes/cs201/Projects/autonomous-weapons/articles/starwars.txt

New Scientist (2002). US ArmySeeks Nanotech Suits. New

Scientist.com; News Service, 14Mar 2002 [online].http://www.newscientist.com/news/news.jsp?id=ns99992043

Pace, S. (1989). MilitaryImplications of Nanotechnology.Foresight Update 6; 1 Aug 1989[online].http://www.foresight.org/Updates/Update06/Update06.2.html

M – P

68

Pergamit, G. and Peterson, C.(1993). Nanotechnology. In:Ringo, T. editor. On The Cutting

Edge of Technology. Indianapolis,IN, USA: Sams Publishing.

Philipson, G. (2001). Human andComputer Merger a Chip of theOld Block. The Age; IT News, 18Sep 2001, p1. For a summary see:http://www.gyre.org/news/related/Metacomputing/Biological+Computing

RAND (2002). RevolutionisingLogistics Support by “ThinkingSmall”. RAND; National Security[online].http://www.rand.org/natsec_area/products/thinkingsmall.html

Reynolds, G.H. (2002). Forward

to the Future: Nanotechnology

and Regulatory Policy. SanFrancisco, CA, USA: PacificResearch Institute.http://www.pacificresearch.org/pub/sab/techno/forward_to_nanotech.pdf

Roco, M.C., and Bainbridge, W.S.,editors (2001). Societal


Nanotechnology. Arlington, VA,USA: National ScienceFoundation.http://www.wtec.org/loyola/nano/NSET.Societal.Implications/nanosi.pdf

Roco, M.C., and Bainbridge, W.S.,editors (2002). Converging

Technologies for Improving

Human Performance.

Nanotechnology, Biotechnology,

Information Technology and

Cognitive Science. Arlington, VA,USA: National ScienceFoundation.http://wtec.org/ConvergingTechnologies/Report/NBIC_pre_publication.pdf

Roman, C. (2002). It’s Ours to

Lose: An Analysis of EU

Nanotechnology Funding and the

Sixth Framework Programme.

Brussels, Belgium: EuropeanNanoBusiness Association.http://www.nanoeurope.org/docs/European%20Nanotech%20Funding.pdf

Roy, R. (2002). Giga Science andSociety. Materials Today;5 (12): 72.http://www.materialstoday.com/pdfs_5_12/opinion.pdf

Ruthfield, S. (1995). The Internet’sHistory and Development. AMC

Crossroads; 2 (1) [online].http://www.acm.org/crossroads/xrds2-1/inet-history.html

Saxl, O. (2000). Opportunities for

Industry in the Application of

Nanotechnology. London, UK:Office of Science and Technology.[A report for The Institute ofNanotechnology, April 2000].http://www.nano.org.uk/contents.htm

Schneider, F.B. and Rodd, M.,editors (2001). International

Review of UK Research in

Computer Science. Swindon, UK:Engineering and Physical SciencesResearch Council.http://www.iee.org/policy/csreport/cs_report.pdf

Schulz, W.G. (2002).Nanotechnology Under the Scope.Chemical and Engineering News ;4 Dec 2002 [online].http://pubs.acs.org/cen/today/dec4.html

Science Blog (2002). New Centreto Transfer Nano Innovations toDefence Industry. Science Blog;10 Dec 2002 [online].http://www.scienceblog.com/community/article553.html

Shim, R. (2002). Sony’s Ando: PCsto Function Like a Brain. ZDNet;News: 5 Dec 2002 [online].http://zdnet.com.com/2100-1105-976269.html

Smalley, R.E. (2001). OfChemistry, Love and Nanobots.Scientific American; 285: 76–7.http://www.ruf.rice.edu/~smalleyg/rick%27s%20publications/SA285-76.pdf

Smith, R.H. (1996). MolecularNanotechnology: Research FundingS o u rces. Nanotechnology Magazine;2 (6).

P – S


Smith, R.H. (2001). Social,Ethical, and LegalImplications ofNanotechnology. In: Roco,M.C., and Bainbridge, W.S.,editors. Societal Implications

of Nanoscience and

Nanotechnology. Arlington,VA, USA: National ScienceFoundation. pp203–11.http://www.wtec.org/loyola/nano/NSET.Societal.Implications/nanosi.pdf

Spooner, J.G. (2002). SonyTeaches Abio New Tricks.ZDNet; News: 10 Oct 2002[online].http://zdnet.com.com/2100-1106-961536.html

Stottler Henke (2002).Glossary of AI Terms.Stottler Henke Associates,

Inc. [online].http://www.shai.com/ai_general/glossary.htm

Thibodeau, P. (2002). SenateScrutinizes US NanotechInvestment. Computer

World; 18 Sep 2002 [online].http://computerworld.com/managementtopics/management/itspending/story/0,10801,74341,00.html

Thompson, H. (1977). ThereWill Always Be AnotherMoonrise: ComputerTechnology and NuclearWeapons. AISB Quarterly;(53/54) [online].http://www.aisb.org.uk/articles/moonrise.html

Trevelyan, J. (1999).Redefining Robotics for theNew Millennium.International Journal of

Robotics Research; 18 (12):1211–23.

Turing, A. (1950).Computing Machinery andIntelligence. Mind; 9:433–60.http://cogprints.ecs.soton.ac.uk/archive/00000499/00/turing.html

Weld, D.S., editor. (1995).The Role of IntelligentSystems in the NationalInformation Infrastructure.American Association for

Artificial Intelligence

[online].http://www.aaai.org/Resources/Policy/nii.html

Whitby, B. (1984) A Code ofProfessionalism for AI. AISB

Quarterly; (64) [online].http://www.aisb.org.uk/articles/professionalism.html

Whitby, B. and Oliver, K.(2001). How to Avoid aRobot Takeover: Politicaland Ethical Choices in theDesign and Introduction ofIntelligent Artifacts. AISB

Quarterly; (104) [online].http://www.aisb.org.uk/articles/avoid_takeover.html

White House Fact Sheet(2000). National

Nanotechnology Initiative:

Leading to the Next

Industrial Revolution.Washington, DC, USA: TheWhite House.http://clinton4.nara.gov/textonly/WH/New/html/20000121_4.html

Wolbring, G. (2002). TheSilenced Ta rgets. In: I n s t i t u t e

of Science in Society Special

Series – Inside Human

Genetics and Genomics; PartII: 28 Jan 2002 [online].h t t p : / / w w w. i -s i s . o rg . u k / S i l e n c e d Ta rg e t s . p h p

Zyvex (2002). Zyvex Web

Site [online].http://www.zyvex.com/Products/tools.html

S – Z

future technologies, today's choices

Documents

Transcript of future technologies, today's choices