Three Laws Blog: Law and Artificial Intelligence

Grieman on Law, Death, and Robots: The Regulation of Artificial Intelligence in High Risk Civil Applications @hartpublishing

Forthcoming from Hart Publishing:

Keri Grieman, Law, Death, and Robots: The Regulation of Artificial Intelligence in High Risk Civil Applications (October 2024). Here from the publisher's website is a description of the book's contents.

 

Can the law keep up with AI? This book examines liability and regulation for artificial intelligence causing serious physical harm, both now and in the future. While AI moves quickly, regulation follows more slowly – an increasing problem for an evolutionary, fast-paced emerging technology. AI has the potential to save lives, but in doing so will have the potential to take them as well. How do we future-proof law and regulation to incentivise life-saving innovation as safely as possible? This book details how to regulate AI in high-risk civil applications (for example, automated vehicles and medicine), addressing both liability and regulatory structure. It highlights crucial liability themes for technology governance; provides tools to bridge the gap between regulators and technologists; examines jurisdictional approaches to AI regulation in the EU, UK, USA, and Singapore; and ultimately suggests a jurisdiction-agnostic blueprint for regulation.

Lee on Synthetic Data and the Future of AI @UCDavisLaw

Peter Lee, University of California, Davis, School of Law, is publishing Synthetic Data and the Future of AI in volume 110 of the Cornell Law Review. Here is the abstract.

The future of artificial intelligence (AI) is synthetic. Several of the most prominent technical and legal challenges of AI derive from the need to amass huge amounts of real-world data to train machine learning (ML) models. Collecting such real-world data can be highly difficult and can threaten privacy, introduce bias in automated decision making, and infringe copyrights on a massive scale. This Article explores the emergence of a seemingly paradoxical technical creation that can mitigate—though not completely eliminate—these concerns: synthetic data. Increasingly, data scientists are using simulated driving environments, fabricated medical records, fake images, and other forms of synthetic data to train ML models. Artificial data, in other words, is being used to train artificial intelligence. Synthetic data offers a host of technical and legal benefits; it promises to radically decrease the cost of obtaining data, sidestep privacy issues, reduce automated discrimination, and avoid copyright infringement. Alongside such promise, however, synthetic data offers perils as well. Deficiencies in the development and deployment of synthetic data can exacerbate the dangers of AI and cause significant social harm. In light of the enormous value and importance of synthetic data, this Article sketches the contours of an innovation ecosystem to promote its robust and responsible development. It identifies three objectives that should guide legal and policy measures shaping the creation of synthetic data: provisioning, disclosure, and democratization. Ideally, such an ecosystem should incentivize the generation of high-quality synthetic data, encourage disclosure of both synthetic data and processes for generating it, and promote multiple sources of innovation. This Article then examines a suite of “innovation mechanisms” that can advance these objectives, ranging from open source production to proprietary approaches based on patents, trade secrets, and copyrights. Throughout, it suggests policy and doctrinal reforms to enhance innovation, transparency, and democratic access to synthetic data. Just as AI will have enormous legal implications, law and policy can play a central role in shaping the future of AI.

Download the article from SSRN at the link.

Diamantis on Reasonable AI: A Negligence Standard @ProfDiamantis @IowaLawSchool @VandLRev

Mihailis Diamantis, University of Iowa College of Law, is publishing Reasonable AI: A Negligence Standard in volume 77 of the Vanderbilt Law Review. Here is the abstract.

Even as artificial intelligence promises to turbocharge social and economic progress, its human costs are becoming apparent. By design, AI behaves in unexpected ways. That is how it finds unanticipated solutions to complex problems. But unpredictability also means that AI will sometimes harm us. To curtail these harms, scholars and lawmakers have proposed strict regulations for firms developing safe algorithms and strict corporate liability for injuries that nonetheless occur. These rigid “solutions” go too far. They dampen innovation and disadvantage domestic firms in the international technology race. The law needs a more nuanced framework that balances progress with fairness. Tort law offers a compelling template, but the challenge is to adapt its distinctly human notion of fault to algorithms. Tort law’s central liability standard is negligence, which compares the defendant's behavior to other "reasonable" people's behavior. But there is no clear comparison class for AI. Assessing algorithms by reference to people would set too low of a bar—AI can and should outperform reasonable humans on many tasks. Assessing AI instead by reference to itself is often impossible—there are not enough algorithms in many contexts to establish a meaningful baseline. This Paper offers a novel negligence standard for AI. Rather than compare any given AI to humans or to other algorithms, the law should compare it to both. By this hybrid measure, an algorithm would be deemed negligent if it causes injury more frequently than the combined incident rate for all actors—both human and AI—engaged in the same type of conduct. This negligence standard has three attractive features. First, it offers a baseline even when there are very few comparable algorithms. Second, it incentivizes firms to release all and only algorithms that make us safer overall. Third, the standard evolves over time, demanding more of AI as algorithms improve.

Download the article from SSRN at the link.

Greenleaf, Mowbray, and Chung on Regulation of (Generative) AI Requires Continuous Oversight (AustLII Submission on the "Safe and Responsible AI in Australia" Discussion Paper) @grahamgreenleaf

Graham Greenleaf, University of New South Wales, Faculty of Law, Andrew Mowbray, University of Technology Sydney, Faculty of Law, and Philip Chung, University of New South Wales, Faculty of Law and Justice, have published Regulation of (Generative) AI Requires Continuous Oversight (AustLII Submission on the ‘Safe and Responsible AI in Australia’ Discussion Paper). Here is the abstract.

Australia’s Department of Industry, Science and Resources invited interested parties to make submissions on a Discussion Paper (DP) Safe and Responsible AI in Australia. This submission by researchers from the Australasian Legal Information Institute (AustLII) addresses the most important general issues identified in the Discussion Paper and suggests the best strategies to address them. In the paper, we make submission on he following: 1 Why the rate of take-up of automated decision-making (ADM) systems in Australia, needs to be determined. 2 Whether regulation should be be aimed at ‘AI generally’, or something else. 3 Why the definition of ‘AI’ should refer to ‘without explicit programming', and why ‘hallucinations should be dropped. 4 An alternative set of 10 principles to be used to guide Australian regulation. 5 The need for a continuous source of expert advice which will regularly report to existing regulatory bodies, to government and to the public. 6 How an ‘Australian Advisory Board on Regulation of AI’ (the AI Board) could be structured. 7 Why Australia should aim to provide inputs to influence international developments where possible, and where this is most important. 8 Why a risk-based approach to regulation of AI should be adopted, 9 Why only the most dangerous applications should be brought within it in the first instance. 10 What could be included in the Commonwealth’s initial ‘AI Framework Act’.

Download the article from SSRN at the link.

Chesterman on From Ethics to Law: Why, When, and How to Regulate AI @ProfChesterman @nus_law @Elgar_Law

Simon Chesterman, National University of Singapore Faculty of Law, is publishing From Ethics to Law: Why, When, and How to Regulate AI in The Handbook of the Ethics of AI (David J. Gunkel, ed., Edward Elgar Publishing, forthcoming). Here is the abstract.

The past decade has seen a proliferation of guides, frameworks, and principles put forward by states, industry, inter- and non-governmental organizations to address matters of AI ethics. These diverse efforts have led to a broad consensus on what norms might govern AI. Far less energy has gone into determining how these might be implemented — or if they are even necessary. This chapter focuses on the intersection of ethics and law, in particular discussing why regulation is necessary, when regulatory changes should be made, and how it might work in practice. Two specific areas for law reform address the weaponization and victimization of AI. Regulations aimed at general AI are particularly difficult in that they confront many ‘unknown unknowns’, but the threat of uncontrollable or uncontainable AI became more widely discussed with the spread of large language models such as ChatGPT in 2023. Additionally, however, there will be a need to prohibit some conduct in which increasingly lifelike machines are the victims — comparable, perhaps, to animal cruelty laws.

Download the essay from SSRN at the link.

Banchio on Reflections on a Legal Framework for Artificial Intelligence

Pablo Rafael Banchio, Facultad Interamericana de Ciencias Sociales (FICS), is publishing Reflections on a Legal Framework for Artificial Intelligence in Revista Doctrina Jurídica; Rivista Giurisprudenza e Diritto Comparato. Here is the abstract.

Objective: Dedicated to the preliminary exposition of the specificity of the legal regulation of the use and development of artificial intelligence, the analysis of various approaches to its definition and the peculiarities of the existing legislative initiatives to enshrine a comprehensive normative regulation of these technologies. Methods: Qualitative and with explanatory scope. Qualitative of the phenomenon to be studied in a comprehensive manner because it allows us to understand the complex world of artificial intelligence from the perspective of law. Explanatory because it focuses on explaining the phenomenon of virtual assistants and the need for regulatory capture; under what conditions it manifests itself, how it is created and how it gives meaning centered on people's lives. Other methods used are formal-logical and comparative-legal. Results: Conclusions are formulated on how to shape legislative support for the use and development of artificial intelligence in a coherent way, considering the specificity of its specific fields of application and balancing the interests of individuals, society, and the state. To respect adequate security and protection of individual rights and interests associated with the development of innovation for the benefit of society, enshrining a supreme principle of justice through universal, comprehensive, and anticipatory regulation. Scientific novelty: There is currently a global problem of an almost total absence of normative regulation, both legal and technical, of the foundations, conditions and peculiarities of the development, launch, operation, integration into other systems and control of the application of artificial intelligence technologies. Practical significance: Artificial intelligence is currently developing intensively, and we are participants in the gradual incorporation of virtual assistants into our lives to facilitate daily activities. An appropriate legal framework will facilitate this insertion, avoid risks associated with the new technologies under development and ensure that individual rights are respected and enforced.

Download the article from SSRN at the link.

Kaminski on Regulating the Risks of AI @MargotKaminski @BULawReview

Margot E. Kaminski, University of Colorado Law School; Yale Information Society Project; University of Colorado, Boulder, Silicon Flatirons Center for Law, Technology, and Enterprise, is publishing Regulating the Risks of AI in volume 103 of the Boston University Law Review. Here is the abstract.

Companies and governments now use Artificial Intelligence (AI) in a wide range of settings. But using AI leads to well-known risks—that is, not yet realized but potentially catastrophic future harms that arguably present challenges for a traditional liability model. It is thus unsurprising that lawmakers in both the United States and the European Union (EU) have turned to the tools of risk regulation for governing AI systems. This Article observes that constructing AI harms as risks is a choice with consequences. Risk regulation comes with its own policy baggage: a set of tools and troubles that have emerged in other fields. Moreover, there are at least four models for risk regulation, each with divergent goals and methods. Emerging conflicts over AI risk regulation illustrate the tensions that emerge when regulators employ one model of risk regulation, while stakeholders call for another. This Article is the first to examine and compare a number of recently proposed and enacted AI risk regulation regimes. It asks whether risk regulation is, in fact, the right approach. It closes with suggestions for addressing two types of shortcomings: failures to consider other tools in the risk regulation toolkit (including conditional licensing, liability, and design mandates), and shortcomings that stem from the nature of risk regulation itself (including the inherent difficulties of non-quantifiable harms, and the dearth of mechanisms for public or stakeholder input).

Download the article from SSRN at the link.

Brownsword on We the Robots? Regulating Artificial Intelligence and the Limits of the Law by Simon Chesterman (Book Review) @KCL_Law @ProfChesterman

Roger Brownsword, King's College London School of Law, has published Book Review: We the Robots? Regulating Artificial Intelligence and the Limits of the Law by Simon Chesterman at 2022 Singapore Journal of Legal Studies 252. Here is the abstract.

Debates about the law, regulation, and governance of robots, artificial intelligence ('AI'), and machine learning have been dominated by voices from North America, Europe, and Australia. While Singapore has hosted conferences at which these matters have been discussed, and while it is clearly an interested party in relation to these debates, a distinctively Singaporean voice has yet to be heard—or, at any rate, it had not been heard prior to the publication of Simon Chesterman’s stimulating book.

Download the review from SSRN at the link.

Craig on The Relational Robot: A Normative Lens for AI Legal Normality (Review of Ryan Abbott, The Reasonable Robot) @craigcarys @DrRyanAbbott

Carys Craig, Osgoode Hall Law School, is publishing The Relational Robot: A Normative Lens for AI Legal Neutrality (Reviewing Ryan Abbott, The Reasonable Robot, Cambridge University Press, 2020) in the Jerusalem Review of Legal Studies (2022). Here is the abstract.

In his impactful book, "The Reasonable Robot" (Cambridge UP, 2020), Ryan Abbott proposes a new guiding tenet for the law’s regulation of Artificial Intelligence: AI legal neutrality. This would establish, as a principled starting point, the default position that “the law should not discriminate between AI and human behaviour.” In this Review Essay, I suggest that AI legal neutrality, as conceived by Abbott, is an interesting proposition but a potentially dangerous default principle with which to equip law for the emerging realities of AI. When we scratch beneath the surface, this concept of neutrality or equal treatment is too focused on the individual person/thing, and too reliant on analogical reasoning and false equivalence. It is therefore too far removed from the normative underpinnings of law, its subjects, and its teleology. Instead, I argue, we need a more substantive notion of law’s technological neutrality—one that looks to the law’s normative objectives to set the default. The place to start, I suggest, is not with a legal neutrality that, by design, disregards the inherent differences between humans and robots and their respective behaviors; on the contrary, we need to be fully attentive to the dynamics of human-robot relations in social context – and alert to the dangers of overlooking differences. Rights and responsibilities should be allocated with a clear view to the relationships and subjectivities they shape and the social values they advance. In Part II, I outline Abbott’s argument in respect of legal neutrality and its application to intellectual property law. I then compare its implications in the patent law sphere to its potential significance in the copyright context. In Part III, I sketch an alternative approach to understanding technological neutrality and law, concerned with consistency in pursuit of normative objectives rather than formal non-discrimination between technologies. I conclude by proposing a relational approach to regulating robots, which would, I believe, bring a much-needed normative lens to the pursuit of legal neutrality.

Download the review from SSRN at the link.

Kop on the EU Artificial Intelligence Act: The European Approach to AI @AIRechtNL @StanfordLaw

Mauritz Kop, Stanford Law School; AIRecht, has published EU Artificial Intelligence Act: The European Approach to AI. Here is the abstract.

On 21 April 2021, the European Commission presented the Artificial Intelligence Act. This Stanford Law School contribution lists the main points of the proposed regulatory framework for AI. The draft regulation seeks to codify the high standards of the EU trustworthy AI paradigm. It sets out core horizontal rules for the development, trade and use of AI-driven products, services and systems within the territory of the EU, that apply to all industries. The EU AI Act introduces a sophisticated ‘product safety regime’ constructed around a set of 4 risk categories. It imposes requirements for market entrance and certification of High-Risk AI Systems through a mandatory CE-marking procedure. This pre-market conformity regime also applies to machine learning training, testing and validation datasets. The AI Act draft combines a risk-based approach based on the pyramid of criticality, with a modern, layered enforcement mechanism. This means that as risk increases, stricter rules apply. Applications with an unacceptable risk are banned. Fines for violation of the rules can be up to 6% of global turnover for companies. The EC aims to prevent the rules from stifling innovation and hindering the creation of a flourishing AI ecosystem in Europe, by introducing legal sandboxes that afford breathing room to AI developers. The new European rules will forever change the way AI is formed. Pursuing trustworthy AI by design seems like a sensible strategy, wherever you are in the world.

Download the article from SSRN at the link.