Bishop on Generative.AI

Lea Bishop (Yale U Yale Information Society Project) has posted “Generative.AI” on SSRN. Here is the abstract:

To expand our collective imagination about the potential, limits, and risks of artificial intelligence, we cannot simply read about it. We must experience it first-hand. This premise makes Generative different from anything else you have read about artificial intelligence.

In an entertaining and thought-provoking package, Generative departs from the mainstream narratives of hype and fear to challenge readers’ assumptions about what ChatGPT and its cousins can (and cannot) do. Through a series of playful Socratic dialogues with actual AI chatbots, Generative vividly demonstrates the technology’s creative power.

Readers will come away with new insights on topics like plagiarism, critical thinking, innovation, education, and the future of artificial intelligence. Like the original Dialogues of Socrates and Plato, Generative’s powerful model of critical inquiry will inspire readers to question our assumptions about artificial intelligence… and everything else.

Conley on Understanding the Duty of Competence for Attorneys Using Generative AI

Anna Conley (U Montana Alexander Blewett III Law) has posted “Understanding the Duty of Competence for Attorneys Using Generative AI” on SSRN. Here is the abstract:

Ethical duties appear poised to be the primary regulatory tool for responsible use of generative AI (“GAI”) by attorneys. This reality necessitates a clear understanding of what the duty of competence requires for attorneys using GAI. Recent state bar and American Bar Association (“ABA”) guidance have coalesced around a foundational concept of informed decision-making, which requires that attorneys have sufficient knowledge about the GAI tool they are using and the specific task at hand to make an informed decision that employing the tool for that task is in the client’s best interests. Competence also requires attorneys avoid automation bias and mitigate against GAI tools’ limitations, including not only hallucinations but also incomplete, inaccurate and misgrounded outputs. Recent ethical guidance requires that attorneys retain cognitive agency when completing tasks that require human judgment and reasoning. These tasks render GAI’s “black box” or lack of explainability, particularly problematic and require attorneys to resist automation complacency, which refers to humans’ reduced capacity to understand or complete the tasks for which they rely on GAI.

Susser & Seeman on Critical Provocations for Synthetic Data

Daniel Susser (Cornell U) and Jeremy Seeman (U Michigan) have posted “Critical Provocations for Synthetic Data” (Surveillance & Society, volume 22, issue 4, 2024[10.24908/ss.v22i4.18335]) on SSRN. Here is the abstract:

Training artificial intelligence (AI) systems requires vast quantities of data, and AI developers face a variety of barriers to accessing the information they need. Synthetic data has captured researchers’ and industry’s imagination as a potential solution to this problem. While some of the enthusiasm for synthetic data may be warranted, in this short paper we offer critical counterweight to simplistic narratives that position synthetic data as a cost-free solution to every data-access challenge—provocations highlighting ethical, political, and governance issues the use of synthetic data can create. We question the idea that synthetic data, by its nature, is exempt from privacy and related ethical concerns. We caution that framing synthetic data in binary opposition to “real” measurement data could subtly shift the normative standards to which data collectors and processors are held. And we argue that by promising to divorce data from its constituents—the people it represents and impacts—synthetic data could create new obstacles to democratic data governance.

Luan et al. on Algorithmic Bias and Physician Liability

Shujie Luan (Johns Hopkins U Carey Business) et al. have posted “Algorithmic Bias and Physician Liability” on SSRN. Here is the abstract:

With the growing use of artificial intelligence (AI) in clinical decision-making, concerns about bias—manifested as differences in algorithmic accuracy across patient groups—have intensified. In response, the U.S. Centers for Medicare and Medicaid Services (CMS) has introduced a liability rule that penalizes healthcare providers who rely on biased algorithms that result in erroneous decisions. This paper examines the impact of this anti-bias liability rule on an AI firm’s development decision as well as a healthcare provider’s decision to use AI. The AI firm develops an algorithm that serves two patient groups, where achieving the same level of accuracy for the disadvantaged group is more costly. The provider then decides whether and how to use AI to make treatment decisions, balancing the reduction in clinical uncertainty against the risk of incurring anti-bias liability. We find the liability rule may induce biased use of AI: The provider may underuse AI overall and disproportionately disregard AI’s recommendations for disadvantaged patients. Interestingly, the effect of liability on AI use is non-monotone: as liability increases, the provider is first less likely to use AI for disadvantaged patients, but then more likely to rely on it. Furthermore, mandating equal algorithmic accuracy across patient groups may inadvertently harm all patients, in part because such mandates may lead to overusing AI for disadvantaged patients.

Bietti on Data is Infrastructure

Elettra Bietti (Northeastern U Law) has posted “Data is Infrastructure” (Theoretical Inquiries in Law (forthcoming 2024)) on SSRN. Here is the abstract:

Data is a contextual phenomenon. It reflects the social and material context from which it is derived and in which it is generated. It embeds the purposes, assumptions and rationales of those who produce, collect, use, share and monetize it. In the AI and digital platform economy, data’s role is primarily infrastructural. Its core uses are internal to companies. Data only rarely serves as a medium of exchange or commodity, and more frequently serves to profile users, train models, produce predictions, bundle and extend product capabilities which in turn are sold to advertisers and other customers. Insofar as they focus on the former, many technical, economic and legal attempts at defining data have inspired reductive policy efforts that include data protection, data ownership and limited data sharing remedies. This paper argues that understanding data as part of infrastructural pipelines can have significant conceptual and policy implications, and can redirect the way privacy, property and antitrust experts understand and govern data. This argument becomes more salient as market actors and regulators grapple with the catalyzing effects of neural networks and generative AI models on digital markets. In antitrust and competition law especially, regulators are consciously adopting a view of data as an infrastructural input into AI and other digital markets. Treating data as an input over which certain firms have competitive advantages can have significant implications for nascent AI markets, and yet the views in antitrust remain too narrow. Understanding data infrastructurally means viewing it not only as a critical input but also as inseparable from other material digital resources such as protocols, algorithms, semiconductors, and platform interfaces; as having important collective functions; and as calling for public interest regulation. Understanding data as infrastructure can move us past limited legal efforts and remedial solutions such as data separations, data sharing, and individual controls, and help reorient how data is produced, stored and managed toward public uses.

Zhang on Nirvana AI Governance: How AI Policymaking Is Committing Three Old Fallacies

Jiawei Zhang (Technische U München (TUM) TUM Social Sciences and Technology) has posted “Nirvana AI Governance: How AI Policymaking Is Committing Three Old Fallacies” (Forthcoming in 15 The Regulatory Review In Depth (2025)) on SSRN. Here is the abstract:

This research applies Harold Demsetz’s concept of the nirvana approach to the realm of AI governance and debunks three common fallacies in various AI policy proposals-“the grass is always greener on the other side,” “free lunch,” and “the people could be different.” Through this, I expose fundamental flaws in the current AI regulatory proposal. First, some commentators intuitively believe that people are more reliable than machines and that government works better in risk control than companies’ self-regulation, but they do not fully compare the differences between the status quo and the proposed replacements. Second, when proposing some regulatory tools, some policymakers and researchers do not realize and even gloss over the fact that harms and costs are also inherent in their proposals. Third, some policy proposals are initiated based on a false comparison between the AI-driven world, where AI does lead to some risks, and an entirely idealized world, where no risk exists at all. However, the appropriate approach is to compare the world where AI causes risks to the real world where risks are everywhere, but people can live well with these risks. The prevalence of these fallacies in AI governance underscores a broader issue: the tendency to idealize potential solutions without fully considering their realworld implications. This idealization can lead to regulatory proposals that are not only impractical but potentially harmful to innovation and societal progress.

Krause on Addressing the Challenges of Auditing and Testing for AI Bias: A Comparative Analysis of Regulatory Frameworks

David Krause (Marquette U) has posted “Addressing the Challenges of Auditing and Testing for AI Bias: A Comparative Analysis of Regulatory Frameworks” on SSRN. Here is the abstract:

The widespread adoption of artificial intelligence (AI) has ushered in a new era of technological innovation, offering transformative benefits in problem-solving and operational efficiency across diverse sectors. However, as AI systems increasingly influence high-stakes decisions, issues of bias and fairness have emerged as critical ethical concerns. This paper explores the multifaceted nature of AI bias-including algorithmic, data-driven, and societal biases-and its pervasive impacts on individuals and communities. Through a comparative analysis of regulatory frameworks for AI bias testing across jurisdictions, this study identifies shared challenges, best practices, and opportunities for improvement. The findings underscore the need for robust regulatory standards that uphold ethical principles in AI use and ensure credible assessments of fairness through third-party audits. The paper proposes targeted recommendations to enhance existing frameworks and suggests new strategies to strengthen the transparency, reliability, and effectiveness of AI bias testing, ultimately supporting a more ethical and accountable AI landscape.

Goodman on AI Accountability Policy Report (NTIA, U.S. Commerce Dept.)

Ellen P. Goodman (Rutgers Law) has posted “AI Accountability Policy Report (NTIA, U.S. Commerce Dept.)” on SSRN. Here is the abstract:

This report was authored for and issued by the U.S. Commerce Department’s National Telecommunications and Information Administration. It calls for improved transparency into AI systems, independent evaluations to verify the claims made about these systems, and consequences for imposing unacceptable risks or making unfounded claims. It makes eight sets of policy recommendations for the U.S. government:

Guidance for 1. AI audits and auditors; 2. AI system and model disclosure; and 3. AI liability standards.

Support for 4. people and tools, including establishing and funding a National AI Research Resource; and 5. research, including into AI system capability and limitation assessments and performance evaluations.

Regulations governing 6. required AI audits and other independent evaluations of high-risk AI model classes and systems; 7. strengthened government capacity across sectors, including maintaining registries of high-risk AI deployments, AI adverse incidents, and AI system audits; and 8. federal contracting to ensure that government suppliers, contractors, and grantees adopt sound AI governance and assurance practices.

Schwarcz et al. on Regulating Robo-advisors in an Age of Generative Artificial Intelligence

Daniel Schwarcz (U Minnesota Law) et al. have posted “Regulating Robo-advisors in an Age of Generative Artificial Intelligence” (Washington and Lee Law Review (2025), Forthcoming) on SSRN. Here is the abstract:

New generative Artificial Intelligence (AI) tools can increasingly engage in personalized, sustained and natural conversations with users. This technology has the capacity to reshape the financial services industry, making customized expert financial advice broadly available to consumers. However, AI’s ability to convincingly mimic human financial advisors also creates significant risks of large-scale financial misconduct. Which of these possibilities becomes reality will depend largely on the legal and regulatory rules governing “robo-advisors” that supply fully automated financial advice to consumers. This Article consequently critically examines this evolving regulatory landscape, arguing that current U.S. rules fail to adequately limit the risk that robo-advisors powered by generative AI will convince large numbers of consumers to purchase costly and inappropriate financial products and services. Drawing on general principles of consumer financial regulation and the EU’s recently enacted AI Act, the Article proposes addressing this deficiency through a dual regulatory approach: a licensing requirement for robo-advisors that use generative AI to help match consumers with financial products or services, and heightened ex post duties of care and loyalty for all robo-advisors. This framework seeks to appropriately balance the transformative potential of generative AI to deliver accessible financial advice with the risk that this emerging technology may significantly amplify the provision of conflicted or inaccurate advice.

O’Grady & O’Grady on Agentic Workflows in the Practice of Law—AI Agents as Ethics Counsel

Catherine Gage O’grady (U Arizona James E. Rogers College Law) and Casey O’grady (Harvard U Harvard Law) have posted “Agentic Workflows in the Practice of Law—AI Agents as Ethics Counsel” (Agentic Workflows in the Practice of Law—AI Agents as Ethics Counsel, 39 Geo. J. Legal Ethics (forthcoming 2025)) on SSRN. Here is the abstract:

Generative AI is reshaping legal practice as law firms invest in AI technology and prepare for a future where AI agents operate alongside human lawyers. While such a future raises numerous ethical concerns, it also opens new opportunities for ethical practice. This article explores the potential for AI agents to improve ethical decision-making within legal practices. We start by defining key concepts such as AI agents and agentic workflows. We then provide a brief overview of the role of ethics counsel in law firms and discuss critical behavioral challenges in the human practice of law. Finally, we introduce a new model of AI agents acting as dedicated ethics counsel. These agents should be specialized, accountable, and systematic to provide comprehensive ethical guidance in a firm’s legal workflows. These ethical agents have the potential to mitigate the human biases in traditional legal practice and offer an efficient and scalable approach to ensure ethical compliance in an increasingly AI-driven legal landscape. To demonstrate this model, we end with real-world examples of what initial ethical agent contributions could look like in practice.