Ethics and AI Innovation ^[1]

Cordel Green
Executive Director, Broadcasting Commission of Jamaica, 9 Central Avenue, Kingston 10, Jamaica
+1 876 929 1998, cgreen@broadcom.org, Secretariat (broadcastingcommission.org)

Professor Anthony Clayton
Chairman, Broadcasting Commission of Jamaica, 9 Central Avenue, Kingston 10, Jamaica
+1 876 929 1998, anthony.clayton@uwimona.edu.jm, Commissioners (broadcastingcommission.org)

Abstract:

1. Introduction

The roll-out of AI applications has been accelerated by the outbreak of theCovid-19 pandemic. The potential contribution of AI in responding to apandemic has been clear for some time; in 2012 the World Economic Forumnoted that "by analyzing patterns from mobile phone usage…we[could]…predict the magnitude of a disease outbreak halfway aroundthe world, allowing aid agencies to get a head start on mobilizingresources and therefore saving many more lives." Most countries failed torealize the significance of this point, however, and the outbreak of theCovid-19 pandemic left them scrambling to develop track and traceapplications.

There is now a much wider understanding of the key role of advancedtechnologies such as informatics and AI in delivering solutions for themanagement of pandemics, including tracking possibly infected individuals,contact tracing, the targeted delivery of healthcare and the ability tolink across databases to elicit important patterns (such as health statusand recent travel history). Clearly, this approach can be effective. Astudy by Oxford University in April 2020 found that if just 56% of acountry's population used a tracking app, it could largely contain theCovid-19 epidemic.

The problem, however, is that this approach raises concerns over privacy,which is why it has had a mixed reception in Western democracies. Oneparticularly important concern is whether personal information is storedexternally rather than on the person's phone. China mandated the use ofelectronic barcodes to store a person's travel and health history, whichplayed a part in helping them to curtail the spread of the virus, and thensuggested at the G20 summit in November 2020 that other countries needed toadopt a uniform set of policies and standards. It is clear that theapproach has some technical merits, but the public reaction in most Westerndemocracies was largely negative, driven by the perception that China wouldthen seek access to everyone's personal data. So, it is important to takeaccount of both the technical feasibility and the social acceptability ofparticular approaches.

The economic choices are equally important. The pandemic caused anastonishingly rapid migration to online teaching and learning, working,meeting and conferencing, administration, shopping and socializing. News,information, entertainment, medical advice and almost all other servicesmoved largely online. The change is probably now irreversible, as manybusinesses, government agencies, universities, retailers and individualshave experienced the efficiency gains and cost reductions of a far moredistributed way of operating.

AI technologies have already revolutionized many fields with applicationssuch as the mass delivery of customized learning experiences, support forthose with visual or other impairments, speech recognition, translationservices, powerful search facilities and personalization of the onlineenvironment, and AI appears set to completely transform industries such asagriculture, manufacturing, shipping, logistics, public and privatetransport, construction, mining, education and many others. The integrationof informatics, AI, robotics, nanotechnology, molecular engineering,biotechnology and others is underpinning the Fourth Industrial Revolution,which is now driving a transformation of social and economic systems thatis "happening ten times faster and at 300 times the scale, or roughly 3,000times the impact" of the first Industrial Revolution (McKinsey GlobalInstitute).

The fourth industrial revolution has astonishing potential and could solvemany of humanity's current problems. However, as David Leslie of the AlanTuring Institute observes:

As with any new and rapidly evolving technology, a steep learning curve means that mistakes and miscalculations will be made and that both unanticipated and harmful impacts will inevitably occur. AI is no exception.^[2]

2. The Challenge

Three challenges are particularly salient. Two of them were addressed byJohn Hopcroft, Turing Award Winner, speaking at the World AI Conference2020 in Shanghai, who said that we have been accustomed to decision-makingby humans or computers, following defined rules, but computers in thefuture will make decisions based on their own learned experience,originating in but not bound by the defined rules in the startingcondition. He also pointed out that goods and services will be produced infuture by a shrinking fraction of the population, which will create anenormous challenge in finding productive, rewarding and remunerated rolesfor the rest of humanity. All industrial revolutions have created far morejobs than they destroyed, but all previous industrial revolutions happenedover far longer periods, allowing more time for adjustment. At present,however, there are signs that new jobs are not being created at the same pace. ^[3]

The third problem is that as the population shifts to rely primarily ononline sources, they become more susceptible to harmful content. Part ofthis is obvious; racism, extreme pornography, conspiracy theories,incitements to violence and radicalization propaganda. But, part of this ismuch more subtle, and includes the way that AI algorithms segregatehumanity into 'bubbles' where dissenting views are no longer heard. Overtime, this can undermine the basis for shared values and tolerance in asociety, and threaten democracy itself.

The World Commission on the Ethics of Scientific Knowledge and Technology(COMEST) has called attention to AI's role in the selection of informationand news that people read, the music that people listen to, the decisionspeople make as well as their political interaction and engagement. Justbefore the pandemic, the UN Secretary General's High-Level Panel on DigitalCo-operation observed that we are increasingly delegating more decisions tointelligent systems, from how to get to work to what to eat for dinner.Underlying these statements is a concern that the AI systems used bytechnology companies are 'black boxes', which open an information chasmbetween the companies and everybody else, including policymakers andregulators. Information is being created, amassed and distributed on anunprecedented scale, but most people have no knowledge of when, the natureor extent to which information about them is being stored, accessed andshared. This gap is one of the most pressing concerns in our transition toa world in which people are developing deeper and closer relationships oftrust with 'smart' devices that are controlled by artificial intelligence.

A related problem is that most people who interact with the AI that liesbehind their apps do so unknowingly. The general willingness to trust theintegrity of providers has allowed the less scrupulous to scrape vastamounts of valuable data that can then be used for marketing or even tomanipulate people's behaviour and choices. Most people don't know that their personal data is someone else's currency. ^[4] In fact, the selling point of the G-MAFIA ^[5] andother technology platforms is that they are proving a wonderful freeservice, allowing unprecedented consumer choice; however, they are alsoselling the consumers to advertisers, as well as selling space on theirplatform to retailers. When the Internet of All Things (IoT) is fullyrealized, devices such as cars, refrigerators, stoves, beds and smarttoilets will also be generating data on their users, leaving the consumerentirely naked in a mass surveillance 'goldfish bowl' society.

Governments, regulators and civil society groups are increasingly focusedon the consequences of the disproportionate power and the potential abuseof influence by social media and big tech, as well as related concernsabout issues such as data privacy, algorithmic bias, disinformation andprofound threats to democracy. Some of the most important emerging conceptsinclude institutional frameworks that can reconcile cross-channeltechnology agnostic regulation with deep-specialist expertise and thedevelopment of new legal concepts of responsibility in the information age,including voluntary or mandated additional obligations to technologyplatform providers to counter and penalize the abuse of social media. It isimportant to realize that threats could include not just conspiracytheorist who encourage violence, but also extend to authoritariangovernments that use their platforms for oppressive and abusive purposesand to spread disinformation, and nationalist leaders who use charges offake news to confuse the public and make it harder to challenge the corruption and fraud in their administrations (Posetti, 2020) ^[6]. Arecent example is the attachment of warning and cautionary labels to postscontaining deliberate untruths by US President Donald Trump before, duringand after the 2020 Presidential election.

Notwithstanding the belated and inconsistent efforts by tech companies toaddress these concerns, the challenges associated with the regulation of AIare formidable for three main reasons:

• First, the pace of technological development now far exceeds theability of most countries to develop the necessary legislative andregulatory frameworks. This is exacerbated by the 'black box' nature of AIsystems and by the fact that genetic algorithms evolve, which makes itharder to devise consistent rules.
• Second, it is difficult to arrive at a regional or internationalconsensus as to the new rules required, because of divergent nationalinterests. For example, the interests of the USA, where most of the majortechnology firms are based, have conflicted with those of the EU withregard to regulation and taxation.
• Third, it is hard to determine the optimal combination of ways tolimit harms while also protecting the consumer's freedom of choice, freedomof expression and personal privacy. This thorny debate is currently focusedon Section 230 of the US Communications Decency Act, which is based on a1996 Congressional policy that sought to promote the unfettered growth ofthe Internet, and grants immunity from liability to social media platformsand other interactive websites. Extensive abuses have made this approachincreasingly untenable, and reform now appears inevitable. The EU's GeneralData Protection Regulation (GDPR) is the most comprehensive solutionproposed to date, but there have been concerns as to whether it willoperate as a form of monetary absolution for big tech, i.e. by allowing (intheory) large technology firms to violate the terms of the GDPR as long asthey regard the gains as worthwhile and the financial sanctions asaffordable. Other measures are possible; in January 2018 Germany imposedpunitive measures on social media companies for allowing unlawful contenton their digital platforms. These measures shift the culpability from theindividual to the platform, with fiscal sanctions if they fail to act. TheUK's Committee on Standards in Public Life recommended a similarlegislative framework that would make social media companies liable forillegal content on their platforms, and in June 2020 the UK's House ofLords Committee on Democracy and Digital technologies recommended thecreation of a regulator to protect democracy by controlling electoralinterference and that technology firms be given a duty of care, withsanctions for firms that fail in their duty (including fines of up to 4% ofglobal turnover or blocking the sites of those found to be seriallynon-compliant).

The challenge, therefore, is to find a way to mitigate the negativeswithout impairing the extraordinary potential of AI for all areas of humandevelopment. AI ethics offers a possible foundation for a more generalizedglobal approach.

3. Ethically-designed AI

Ethics is the conscience of the law. It is aspirational, in that itnormally requires a higher standard of behaviour than the rules of lawcurrently dictate. AI ethics is an ideal of how AI should be, as opposed toa minimum standard to which AI must comply.

The Turing Institute defines AI ethics as 'a set of values, principles, andtechniques that employ widely accepted standards of right and wrong to guide moral conduct in the development and use of AI technologies.' ^[7]This is a human-centric approach to AI, based on "privacy, accountability,safety and security, transparency and explainability, fairness andnon-discrimination, human control of technology, professional responsibility, and promotion of human values." ^[8] Thedefinition may appear simple, but the application is challenging, with anumber of unresolved issues. One key question is whether the appropriatelegal framework for AI is soft or hard law. This can be understood as achoice between self-regulation grounded in internal corporate policy andinternational guidelines on the one hand, and statutory and regulatoryapproaches on the other.

One important indicator of the possible way forward is that soft law isdeveloping rapidly, and there is a growing consensus that ethical normsmust be developed for the governance of AI, although it is likely that thisalso reflects the difficulty of incorporating these norms into hard law.Some principles and declarations do now exist. These include thepublication of Ethics Guidelines for Trustworthy AI by the EuropeanCommission's High-Level Expert Group on Artificial Intelligence; UNESCO iscurrently conducting global consultations on recommendations that have beendeveloped by an expert group, and the UN SG also established a High-LevelPanel which has produced a report. Some large enterprises have alsopublished their own AI ethics principles. The G7 recently announced aglobal partnership on AI (GPAI) to support and guide the responsibledevelopment of artificial intelligence that is grounded in human rights,inclusion, diversity, innovation, and economic growth, and GPAI's expertswill also investigate how AI can be leveraged to better respond to andrecover from COVID-19.

One widely-held view, at least in the private sector, is that industryself-regulation is best suited for the rapid speed at which AI isdeveloped, the assumption being that such regulation will be faster andmore agile than regulatory bodies that are established by government. Theexperience, though, is that the 'soft law' systems that have beenestablished at the company level have been found badly wanting, and arelargely the results of reactive attempts at public relations. Theseself-regulatory processes tend to rely on a high level of automation (particularly with social media), using algorithms ^[9] tosearch vast data sets for problematic material. However, there are a numberof problems with this approach:

• First, there may be concealed bias (Amar, 2019) ^[10].
• Second, algorithms cannot screen entirely autonomously, for anumber of reasons. One is context. In English, for example, words can bemodified by context or intonation and irony can turn a word into theopposite of its nominal meaning. Humans understand context and metaphor,but this is hard to encode. Another that words can be used to signifysomething that is obvious only to initiates.
• A third is that language is fluid; English, for example, is spokenin many dialects and accents, which constantly evolve.
• A fourth is that harmful misinformation can be presented in anacceptable form; spurious information about the dangers of vaccines can bepresented in a pseudo-scientific manner that makes it appear credible (Temperton, 2020. ^[11].
• A fifth is that it may be difficult to define when religion becomespolitical, and when an appeal for spiritual struggle is actually a call forjihad.
• A sixth problem is that terrorists can change platforms and spreaddifferent messages across multiple platforms, and terrorist organizationscan morph into new forms, so that an algorithm may become increasinglyinaccurate unless it is constantly retrained with new material (Ammar,2019).
• A seventh problem is that there is a fundamental conflict betweenthe business model of social media companies, which is based on advertisingwhich is generated by viral content, and the idea that they should excludeposts that generate a lot of traffic.
• An eighth potential problem is that the reliance on technologycompanies to use AI-based algorithms to moderate content amounts to theprivatization of censorship. This would have mattered less in the past, butnow that technology companies are, in effect, by far the largest mediacorporations in the world, it matters a great deal.

So, while algorithms can reduce the problem of volume, they cannot replacethe humans who must be involved in further rounds of screening. However, itis impossible for humans to screen more than a tiny fraction of the volumesof content in social media, so the solution is likely to involve acombination of better algorithms and tiered human screening. This willclearly involve the technology firms, who have the capacity to do this.However, given their largely reactive response to the abuses taking placeon their platforms, many people now feel that tech companies can no longerbe trusted to be the sole arbiters to draw the boundaries and, as thesocial impacts are now very far-reaching, there must be someindependently-determined standards (which almost certainly means governmentregulation). So, there is as yet no common agreement as to how to draw theethical boundaries, or who should draw them, who should apply them, whoshould enforce them and how they should be enforced.

The EU has been far more sanguine about the potential to develop a hard lawapproach. It has introduced the General Data Protection Regulation (EUGDPR) and the European Parliament has called for a central regulatory body,similar to the Food and Drug Administration, to assess the impact of algorithms before they are deployed. ^[12]Hard law approaches must, however, take into account the 'pacing problem',which is that overly restrictive law and regulations can slow down the paceof technological innovation, while also addressing the concern thatdisruptive technologies are currently developing at a far faster pace than policy and regulations can adapt. ^[13]This is an example of Collingridge's dilemma (Collingridge, 1980), whichstates that 'attempting to control a technology is difficult…becauseduring its early stages, when it can be controlled, not enough can be knownabout its harmful social consequences to warrant controlling itsdevelopment; but by the time these consequences are apparent, control has become costly and slow'. ^[14]

There are also intermediate options. The progress that is being made in thedevelopment of soft law may also have a positive influence in shaping the development of hard la. ^[15].Like the campaigns against tobacco and climate change, a grassroots,down-up network of soft proposals and interventions may eventually becodified in a hard legal outcome.

3.1. AI and Legal Responsibility

Further ethical challenges lie ahead. Transhumanist philosophy aspires tothe redesign of humanity to allow us to transcend our biologicallimitations, and to 'shape the human species through the direct application of technology'.^[16] For some, this includes a definition of AI that approximates 'some aspect of human or animal cognition using machines'. ^[17]This implies that at some point in future machines will become sentient,with implications for their claim to have rights and the imposition ofsocial and legal obligations. There are fears that the growing influence ofAI in human affairs could eventually challenge the very concept of beinghuman, and the rights which depend on that status. Although he was writingwith genetics in mind, John Harris' statement is equally true of AI:

[it] is...beginning to create a new generation of acute and subtle dilemmas that will in the new millennium transform the ways in which we think of ourselves and of society... bringing both a new understanding of what we are and almost daily developing new ways of enabling us to influence what we are, that is creating a revolution in thought, and not least in ethics.^[18]

Throughout history, humans have adapted and adjusted to the technologies ofthe time. The integration of AI into all human experience anddecision-making will come to be seen as normal and taken for granted, butthere will still be a number of profound ethical choices that must be made.

A human-centric point of departure is that machines are created by humans,and that the objective of any status accorded to an intelligent machineshould therefore be determined solely by human utility, rather than theinterest of the intelligent machine itself. That is, the purpose of anyright which is extended to or created for an artificial entity should bethat it provides some benefit for humans. Another view is that intelligentmachines should not be conferred with personhood solely on the basis oftheir functional intelligence, or because humans depend on them, and thatmachines cannot be held to human standards even if they are attributed withhuman characteristics such as 'smart' or 'autonomous', or of having agency.Humans have concepts such as accountability, ethics, values and morality,which guide their behaviour. Machines may have working hypotheses, but theydo not have beliefs or moral values, which means that they cannot be heldaccountable for moral lapses. Military robots can be given rules ofengagement that are based on legal and moral values, but they do notexperience suffering or guilt if there is a mistake when applying thoserules and innocents are killed.

However, the law has already granted juridical personhood to a company, soit must be possible that it will be granted to another artificial entitywhich is autonomous and by some measure intelligent, which a company is not. ^[19]For example, who should be responsible for any bad decision made by a fullyautonomous vehicle which is not due to any defect in their manufacturing?One proposal is that responsibility should be attached to the autonomousmachine so that liability is not dependent on its ownership or manufacture.An autonomous vehicle would therefore be required to have its own insurancecoverage. There might not be very many claims, as these machines don't getdrunk, tired or emotional, and are likely to get into far fewer accidents.

Any proposal to extend rights to machines would have to take into accountthe role of the machine, its level of autonomy and intelligence, and theextent to which a human right will be protected by that machine right. Themore integral an intelligent machine is to the preservation and protectionof a human right, and in helping to make human life better and moremeaningful, the more society and our legal system will tend to be disposedto according them some form of legal personality.

Consider some possible scenarios for a society which is increasinglyaccommodating of 'synthetic' experiences:

• If a human has sex with a sex robot that they do not own, shouldthis be treated as interference with property or akin to personal assault,not necessarily because the robot might be self-aware but as a response tothe effect of the action on the robot's owner?
• Sex robots are now being buil. ^[20],and several social and legal issues have already emerged. One is whetherpaedophiles should be allowed to have sex with robots that resemblechildren. Opinions are sharply divided on this point. Some feel that thewhole idea is immoral, others fear that this would encourage the user to goon to try to have sex with real children, but some therapists who work withsex offenders have argued that these devices could become an importantdeterrent to child molestation by allowing the individual to act out theirurges without victimizing anyone.
• If a simulacrum robot operates as its owner's double, and used toact remotely on behalf of its owner, should it be considered an agent forthe purposes of law, with a right to exercise such authority as its humanprincipal confers? Could they have power of attorney with regard to e.g.making important decisions? If the owner were to request euthanasia, couldthe robot grant that request? If the robot breaks the law, is the robotliable, or the owner?
• If a badly-injured human has life-saving surgery, and ends up withmore intelligent and autonomous machine parts than original components,they would almost certainly be considered to still be a human being. But ifa machine were given human biological components, and ended up being morehuman than the original human (at least by weight), would the same ruleapply? If not, then does the starting condition matter more than the endstate?
• Should a cyborg be clothed with personhood, whether juridical orconstitutional, and with or without exceptions? Should a cyborg's rightsshould be constrained in order to protect humanity? What would happen whencyborgs can out-perform humans in most areas?
• Could there come a time at which 'ownership' of robots is no longerseen as acceptable, just as slavery came to be unacceptable?

The privacy of information which is generated by interactions andtransactions with social machines is also an important ethical concern.Savirimuthu raises the question:

While we may not have too much ethical concerns about the use of Roomba helpers or Alexa in domestic settings, is there an ethical line that is crossed when robot sex brothels and voice recognition devices can be used for self-gratification or emotional engagement?^[21]

And also (on p. 342, ibid)

The problematizing of HRI [Human-Robot Interaction] in the emotional/sexual domain also draws attention to the susceptibility of individuals to being manipulated, particularly when data-driven processes become the proxy for constituting and ordering relations, preferences and values often without the user's awareness…The broader point that seems to emerge from each of these contributions in this part may be that engineers and philosophers will need to better understand each other so that steps can be taken to find engineering solutions that correspond with human values and ethical norms. This is a legitimate goal for identifying and developing rules to a point. There is, however, the added dilemma of defining the ethical landscape for robotics since the normative structures (whether Utilitarian or Kantian) are fluid and creating a hierarchy of values not entirely free from their own questions of rights to be prioritized. Where does one actually start when allocating to robots the range of universal rights for robots? How do we avoid the problem of over-or-under inclusion? Who decides and should the robot be given discretion? Can we program robots to behave ethically or must there be a human in the loop?^[22]

3.2. The Importance of Trust

There has been a notable decline in trust in public institutions. In theUSA, for example, a recent Economist/YouGov Poll found that 75% ofregistered voters think that voter fraud occurred during the 2020presidential election (3% of Biden voters and 81% of Trump voters thought that fraud had influenced the outcom. ^[23]), which could seriously undermine faith in democracy in the US. Today,many people are more likely to trust relationships, rather thaninstitutions. This change coincides with the deepening of trusted personalrelationships mediated through social networks that are controlled byalgorithms, and closer relationships of trust with personal smart devices.This means that society will become increasingly exposed to the risk thatalgorithms will further segregate humanity into 'bubbles' where dissentingviews are no longer heard. Over time, this could further undermine thebasis for shared values and tolerance in a society. Alternatively, it wouldbe possible to edit the algorithms to encourage exposure to some critical,dissenting or challenging views. These are not just technological choices;they have profound implications for our future.

4. A Framework for Action

There is an important question as to whether new approaches to regulationor other forms of government intervention are now required, whether atechnological model (i.e. using algorithms to take down problematicmaterial) is now the only viable solution, or whether a hybrid approach(combining, for example, regulation, education and reputational pressure)might have the best chance of success. Many countries are wrestling withthese issues, and different possible models are being developed. It isanalogous to the development of road traffic laws. Every country craftedits own road traffic laws, with different offences and penalties, but everycountry has road traffic laws. In regard to AI, there are some commonprinciples that have emerged, albeit expressed differently. Some of theseare as follows:

1. It should be possible to explain how AI works and what an algorithm is doing.
2. The data used to train AI systems should be transparent and verifiable.
3. Developers and companies should incorporate ethical guidelines when developing autonomous intelligent systems.
4. It should be possible to attribute accountability for AI-driven decisions and the behaviour of AI systems.
5. All citizens must have some idea of what algorithms do and a basic understanding of how AI works.
6. AI should be developed and implemented in accordance with international human rights standards, with an emphasis on strengthening freedom of expression, universal access to information, the quality of journalism, and media pluralism, while mitigating against the spreading of disinformation, including terrorism, violent extremism, hate speech and fake news (although there are important issues of definition here).
7. AI should be aimed to avoid bias and allow for cultural diversity.

Implementing ethical AI will require a multi-modal and co-regulatoryapproach, involving actors across all vectors of information - acrossplatforms, across devices and unrestricted by the physical borders. Theseactors will be policy makers, regulators, operators, content creators,aggregators, intermediaries, users and civil society. For their part,regulators must be evidenced-based and rules must function across platformsin a technologically agnostic manner. This means that regulators must becapable of using ethically-designed AI systems that can be deployed in acomplex media ecosystem. Finally, one of the most important responses tothe challenges and opportunities of AI is digital literacy, so that thateveryone understands the role of algorithms in the AI systems with whichthey interact and the ethical considerations and expectations for thedesign and use of such systems.

5. Conclusion

We now must choose, as Carlos Moreira and David Ferguson observe in their book, 'The transHuman Code. ^[24],between building a better future with the help of technology or building afuture with better technology - at the expense of most of humanity. We willhave to choose whether to live in free countries empowered by technology,or in authoritarian regimes that use technology to control their people. Wewill have to choose between living in a world with rules but no walls, or corralled into pens demarcated by nationalist 'great walls'. ^[25]

This is not the first time that technological innovation has driven socialtransformation. Between 1850 and 1870, for example, the invention ofdynamite, the railway, sewing machines, the laying of the transatlanticcable, improvements in agriculture, and advances in surgery and anesthesiachanged lives and destinies. The same period saw the development oflong-range artillery and modern warfare. Now AI has the potential to be thegreatest liberator or the greatest oppressor of humanity. Humanity hasalways faced choices: we have survived so far. We can only hope that wewill choose our next steps wisely.

6. References

Capurro, Rafael: Agar, N. Don't Worry about Superintelligence. Journal of Evolution and Technology 26 (1) (February): 73-82, 2016. (http://jetpress.org/v26.1/agar.htm)

Agar, N. "Why a technologically enhanced future will not be as good as we think", OUP Blog, posted July 20, 2015. (https://blog.oup.com/2015/07/future-of-technology/)

Agar, N. We must not create beings with moral standing superior to ours, Journal of Medical Ethics 39 (11):709, 2013.

Allan, T and Widdison,R. 'Can computers make contracts?', Harvard Journal of Law and Technology, 9, 25-52, 1996. http://jolt.law.harvard.edu/articles/pdf/v09/09HarvJLTech025.pdf

Bayamlioglu, E. Intelligent Agents and Their Legal Status, 2008. http://www.ankarabarosu.org.tr/siteler/AnkaraBarReview/tekmakale/2008-1/8.pdf

Bernat, E. "Which Beings Should Be Entitled to Human Right?" Medical Law International March, Vol 9 no.1 pp. 1-12, 2008.

Bostrom, N. Superintelligence: Paths, dangers, strategies. Oxford: Oxford University Press, 2014. (http://subvert.pw/a/Superinteligence.pdf)

Bostrom, N. When Machines Outsmart Humans, Futures. Vol. 35:7, 759, 2014.

Bostrom, N. A History of Transhumanist Thought, Academic Writing Across the Disciplines, eds. Michael Rectenwald & Lisa Carl (New York: Pearson Longman, 2011. https://www.nickbostrom.com/papers/history.pdf

Brooks, R.A. 'Intelligence without representations', Artificial Intelligence 47, 139-159, 1991. http://ozark.hendrix.edu/~ferrer/courses/235/f14/readings/brooksIntelligenceWithoutRepresentation.pdf

Brownsword, R. 'What the World Needs Now: Techno-Regulation, Human Rights and Human Dignity' in R. Brownsword (ed.), Global Governance and Human Rights, 2019. http://kavehh.com/my%20Document/KCL/Technology%20law/what%20the%20world%20needs%20now,%20techno-regulation%20human%20rights%20human%20dignity.pdf.

Bushaus, D, "Does the 'G Mafia; control the future of AI" Inform January 2019. https://inform.tmforum.org/insights/2019/01/g-mafia-control-future-ai/

Bygrave, L.A. "Electronic Agents and Privacy: A Cyberspace Odyssey 2001" - Int J Law Info Tech, Vol 9 (3): 275, 2001.

Calo, R. Artificial Intelligence Policy: A Primer and Roadmap, 51 U.C. Davis L. Rev. 399, 2017.

Carroll, J "Artificial Intelligence & Robotics in Construction: "Massively Transformative", January 17, 2019. https://jimcarroll.com/2019/01/article-artificial-intelligence-in-construction-massively-transformative/

Chropra, S & White, L. Artificial Agents - Personhood in Law and Philosophy, 2004. http://www.sci.brooklyn.cuny.edu/~schopra/agentlawsub.pdf

Deng, B. Machine ethics: The robot's dilemma Working out how to build ethical robots is one of the thorniest challenges in artificial intelligence. Nature International Weekly Journal of Science, July 1. 2015 http://www.nature.com/news/machine-ethics-the-robot-s-dilemma-1.17881 .

Fenwick, Mark D.; Kaal, Wulf A. Ph.D., & Vermeulen, Erik P.M. "Regulation Tomorrow: What Happens When Technology Is Faster than the Law?," American University Business Law Review, Vol. 6, No. 3, 2018. Available at: http://digitalcommons.wcl.american.edu/aublr/vol6/iss3/1

Field, C. South Korean Robot Ethics Charter. PhD thesis (part), University of Technology, Sydney. South Korean Robot Ethics Charter, 2012. https://akikok012um1.wordpress.com/south-korean-robot-ethics-charter-2012/ .

Fjeld, J et al "Principled Artificial Intelligence", Berkman Klein Center, February 14, 2020. https://cyber.harvard.edu/publication/2020/principled-ai

Future of Life Institute, Global AI Policy. n.d.: https://futureoflife.org/ai-policy/

Gasser, Urs, & Virgilio A.F. Almeida. "A Layered Model for AI Governance." IEEE Internet Computing 21 (6) (November): 58-62, 2017. doi:10.1109/mic.2017.4180835.

Citable link http://nrs.harvard.edu/urn-3:HUL.InstRepos:34390353

Genus, A & Stirling, A. Collingridge and the dilemma of control: Towards responsible and accountable innovation. Research Policy, 2017. 10.1016/j.respol.2017.09.012. https://www.researchgate.net/publication/320245067_Collingridge_and_the_dilemma_of_control_Towards_responsible_and_accountable_innovation/citation/download

Glenn, L.M. "Biotechnology at the Margins of Personhood: An Evolving Legal Paradigm", Journal of Evolution and Technology, Vol 13, 2003. http://jetpress.org/volume13/glenn.htm

Glenn, L.M. "Case Study: Ethical and Legal Issues in Human Machine Mergers (Or the Cyborgs Cometh), 21 Annals Health L. 175, 2012. Available at:http://lawecommons.luc.edu/annals/vol21/iss1/16

Glen, L.M. "A Legal Perspective on Humanity, Personhood, and Species Boundaries",The American Journal of Bioethics, Volume 3, Number 3, pp. 27-28, 2003. https://muse.jhu.edu/login?auth=0&type=summary&url=/journals/american_journal_of_bioethics/v003/3.3glenn.html

Goodman, J. "The Only Way is Ethics", The Law Society Gazette, 21 May 2018. https://www.lawgazette.co.uk/features/the-only-way-is-ethics/5066162.article

Green, C and Clayton, A; "Tipping or Tripping Point, pp. 68-69, 2019. https://www.iicom.org/feature/communications-tipping-or-tripping-point/

Gunkel, D. "Apocalyptic rhetoric about AI distracts from more immediate, pressing concerns ", December 9, 2014. http://newsroom.niu.edu/2014/12/09/apocalyptic-rhetoric-about-ai-distracts-from-more-immediate-pressing-concerns/

Information Age, "will.i.am on artificial intelligence and data independence", 2018. https://www.information-age.com/will-i-am-artificial-intelligence-123475325/

Karnow,C.E.A. 'The Encrypted Self: Fleshing out the Rights of Electronic Personalities', The John Marshall Journal of Computer & Information Law, pp. 1, 12-13, 1994. http://repository.jmls.edu/cgi/viewcontent.cgi?article=1335&context=jitpl

Leslie, D. Understanding artificial intelligence ethics and safety: A guide for the responsible design and implementation of AI systems in the public sector. The Alan Turing Institute, 2019. https://doi.org/10.5281/zenodo.3240529

Lutz, C, Schottler, M., & Hoffmann, CP. "The privacy implications of social robots: Scoping reviews and expert interviews", Mobile Media & Communication, Vol. 7(3) 412-434, 2019.

McDermott, D. "Why Ethics is a High Hurdle for AI", n.d. http://www.cs.yale.edu/homes/dvm/papers/ethical-machine.pdf

More, M and Vita-More, N. The Transhumanist Reader, 2018. http://www.arise.mae.usp.br/wp-content/uploads/2018/03/The-Transhumanist-Reader-Classical-and-Contemporary-Essays-on-the-Science-Technology-and-Philosophy-of-the-Human-Future_cap1.pdf

Moreira, C and Ferguson, D. The transHuman Code, Greenleaf Book Group press, 2019.

Muehlhauser, L. "Intelligence Explosion and Machine Ethics" (https://intelligence.org/files/IE-ME.pdf)

Negroponte, Being Digital (Hodder & Stoughton: London 1995), especially pp. 149 et seq; referenced in Electronic Agents and Privacy: A Cyberspace Odyssey 2001 - Int J Law Info Tech, Vol 9 (3): 275 Lee A. Bygrave, 2001.

Putnam C, The Doctrine of Man: A Critique of Christian Transhumanism April 28, 2011, https://www.academia.edu/4162109/A_Critique_of_Christian_Transhumanism

Rivard, M.D. Toward a General Theory of Constitutional Personhood: A Theory of Constitutional Personhood for Transgenic Humanoid Species, 39 UCLA Law Review 1425, 1992.

Rivard, M.D. Constitutional Personhood. Power Point presentation at Terasem Movement Symposium, December 10, 2006.

Ruse H.G. Electronic Agents and the Legal Protection of Non-creative Databases. Int J Law Info Tech, Vol 9 (3): 295, 2001.

Savirimuthu, J. Robot Ethics 2.0: From Autonomous Cars to Artificial Intelligence - Int J Law Info Tech, Vol 26 (4): 337 at 342, 2018.

Solum, L.B. Legal Personhood for Artificial Intelligences. North Carolina Law Review, pp. 1231, 1992. Electronic Paper Collection:http://papers.ssrn.com/abstract=1108671

The Guardian, 25 October 2011, https://www.theguardian.com/technology/2011/oct/25/john-mccarthy

UN Secretary-General's High-level Panel on Digital Cooperation, 2018. "the age of digital interdependence Report" https://www.un.org/en/pdfs/DigitalCooperation-report-for%20web.pdf

Weiss, A. Validation of an Evaluation Framework for Human-Robot Interaction. The Impact of Usability, Social Acceptance, User Experience, and Societal Impact on Collaboration with Humanoid Robots. PhD thesis, University of Salzburg, 2010.

Wooldridge & Jennings, 'Intelligent Agents: Theory and Practice'. 10 The Knowledge Engineering Review, no. 2, pp. 115-152, 1995. http://www.cs.ox.ac.uk/people/michael.wooldridge/pubs/ker95.pdf

World Economic Forum, "What is the Fourth Industrial Revolution?", 2016. https://www.weforum.org/agenda/2016/01/what-is-the-fourth-industrial-revolution/ , 2016 WEF, "What is the Fourth Industrial Revolution?" https://www.weforum.org/agenda/2016/01/what-is-the-fourth-industrial-revolution/

World Economic Forum, "The Future of Jobs Report", 18 January 2018. https://www.weforum.org/reports/the-future-of-jobs

The International Review of Information Ethics , Vol 29, 2021