Ryan Whalen (The University of Hong Kong – Faculty of Law) has posted “Defining Legal Technology and Its Implications” (International Journal of Law and Information Technology, 2022) on SSRN. Here is the abstract:

Legal technological developments have been both lauded as the promising future of the law and derided as a danger to the fundamentals of justice. This article helps reconcile these divergent perspectives by providing a definition of legal technology and a framework through which to understand its different types and their potential implications for the legal system and society more generally. Mapping technologies according to how specifically they afford legal uses, and the directness with which they engage in unmediated legal activities reveals different technological categories and their differing propensities to have legal, functional or general implications. This framework can help inform discussions both about which types of legal technologies to be excited about, and which to be concerned about, while also helping guide research, policymaking, design and adoption considerations.

Grzegorz P. Mika (AI Transparency Institute) and Eva Thelisson (same) have posted “Application of Swiss Private Law by AI” on SSRN. Here is the abstract:

The current debate on the deployment of Artificial Intelligence (AI) in the judicial process raises the question of how humans apply the law, and if AI can be of use in this process. Similar to other legal systems, Swiss private law provides for explicit rules and guidance prescriptions as to its own application, which judges have to apply in the decision-making process. These rules for instance mandate the inclusion of meaning, as opposed to wording, of any statute. These rules also prescribe the way of ruling in the absence of statute. These rules command good faith as well as equity to construe and sometimes to limit or deny rights and duties at stake. Good faith in particular governs the rules and principles of interpretation of contracts and other expressions of intent in private law. Equity is meant to serve as guidance to apply openly or broadly formulated statutes. AI would also have to observe these rules and principles of application of the law. This article aims at assessing whether AI systems could comply with these rules of judicial ruling.

Cass R. Sunstein (Harvard Law School) has posted “Welfare Now” (Duke Law Journal, forthcoming) on SSRN. Here is the abstract:

Behaviorally informed interventions include nudges, taxes, subsidies, bans, and mandates. In evaluating such interventions, policymakers should consider both their welfare effects (including, for example, their potentially negative effects on subjective well-being) and their effects on distributive justice (including, for example, their potentially negative effects on those at the bottom of the economic ladder). Preference satisfaction matters to welfare, but preference satisfaction is not foundational: People might prefer Option A over Option B, but if Option B produces more welfare than Option A, we should not celebrate a situation in which everyone ends up with Option A. The arguments for investigating welfare effects, and effects on distributive justice, are meant as objections to efforts to evaluate behaviorally informed interventions solely in terms of (for example) ex ante revealed preferences and effects on participation rates. The arguments are also meant as pleas for analysis of the distributive effects of such interventions and for specification and investigation of their welfare effects, including their effects on experienced well-being. A pervasive concern is that behaviorally informed interventions might have negative welfare effects on subjective well-being that are easily ignored – as, for example, when information disclosure makes people sad or scared, or when a shift to healthier eating makes people enjoy their meals less. At the same time, such interventions might have positive effects on subjective well-being that are easily ignored – as, for example, when information disclosure makes people feel confident and safe, or when a shift to healthier eating makes people enjoy their meals more.

Ilias Chalkidis (University of Copenhagen; Athens University) et al. have posted “LexGLUE: A Benchmark Dataset for Legal Language Understanding in English” on SSRN. Here is the abstract:

Law, interpretations of law, legal arguments, agreements, etc. are typically expressed in writing, leading to the production of vast corpora of legal text. Their analysis, which is at the center of legal practice, becomes increasingly elaborate as these collections grow in size. Natural language understanding (NLU) technologies can be a valuable tool to support legal practitioners in these endeavors. Their usefulness, however, largely depends on whether current state-of-the-art models can generalize across various tasks in the legal domain. To answer this currently open question, we introduce the Legal General Language Understanding Evaluation (LexGLUE) benchmark, a collection of datasets for evaluating model performance across a diverse set of legal NLU tasks in a standardized way. We also provide an evaluation and analysis of several generic and legal-oriented models demonstrating that the latter consistently offer performance improvements across multiple tasks.

Recommended.

Alexandra Molitorisova (University of Bayreuth, Faculty of Law; Masaryk University), Kai P. Purnhagen
(University of Bayreuth; Erasmus University of Rotterdam – Rotterdam Institute of Law and Economics), and Pavel Šístek have posted “Techno-regulation: Technological Collaboration between EU Administrations” on SSRN. Here is the abstract:

This article examines different forms of technological collaboration between Member States’ public administrations as currently present in the EU, namely institutional and transactional, drawing from examples in two sectors – telecommunications and food. The article argues that different collaboration forms can be explored more systematically by policy makers when faced with techno-regulatory choices. It subsequently argues that when developing techno-regulatory tools for the implementation and enforcement of EU law, national regulatory authorities should place technological cooperation at the forefront of their policy considerations. It concludes with a plea for an increased reciprocity in technological collaboration based on open-source solutions.

Eldar Haber (University of Haifa Law) has posted “Algorithmic Inclusion” (72 Fla. L. Rev. F. 94 (2021)) on SSRN. Here is the abstract:

Artificial Intelligence (AI) is expected to dramatically change humanity. From the automation of daily tasks and labor, to curing diseases and handling disasters, many forecast that human beings will soon begin enjoying the benefits of AI technology within many aspects of their lives. While it is currently difficult to evaluate when and to what extent AI will live up to fulfill its promise, it is uncertain whether the continued development of AI technology will widen the already existing digital divide between those with access to technology and those without.

The concern of a new digital divide that could stem from AI technology had been articulated by Professor Peter K. Yu as the algorithmic divide. In his Article, Professor Yu describes the potential inequalities that these technological developments will likely create and intensify. Much like the digital divide, Professor Yu argues, there will be a “new inequitable gap” between those with access to new technologies and those without, while the latter will miss out “on the many political, social, economic, cultural, educational, and career opportunities provided by machine learning and artificial intelligence.” This Response adds to the discussion of the perceived forthcoming algorithm divide by further analyzing key issues that emerge within the goal of inclusion. The first Part briefly summarizes the algorithmic divide as projected by Professor Yu and his suggestions to reduce the risks and fears that stem from it. The second Part then raises further caveats and key issues that must be taken into consideration when discussing how to bridge the algorithmic divide.

Hannah Bloch-Wehba (Texas A&M University School of Law; Yale ISP) has posted “Algorithmic Governance from the Bottom Up” (Brigham Young University Law Review, Forthcoming) on SSRN. Here is the abstract:

Artificial intelligence and machine learning are both a blessing and a curse for governance. In theory, algorithmic governance makes government more efficient, more accurate, and more fair. But the emergence of automation in governance also rests on public-private collaborations that expand both public and private power, aggravate transparency and accountability gaps, and create significant obstacles for those seeking algorithmic justice. In response, a nascent body of law proposes technocratic policy changes to foster algorithmic accountability, ethics, and transparency.

This Article examines an alternative vision of algorithmic governance, one advanced primarily by social and labor movements instead of technocrats and firms. The use of algorithmic governance in increasingly high-stakes settings has generated an outpouring of activism, advocacy, and resistance. This mobilization draws on the same concerns that animate budding policy responses. But social and labor movements offer an alternative source of constraints on algorithmic governance: direct resistance from the bottom up. These movements confront head-on the entanglement of economic power, racial hierarchy, and government surveillance.

Using three case studies, this Article explores how tech workers and social movements are resisting and mobilizing against technologies that expand surveillance and funnel wealth to the private sector. Each case study illustrates how the intermingling of state and private power has required movements to engage both within and outside firms to counteract the growing appeal of automation. Yet the dominant approaches to regulating the government’s uses of technology continue to afford a privileged role to private firms and elite institutions, sidelining movement demands. The fundamental challenge posed by these movements will be whether—and how—law and policy can accommodate demands for bottom-up control. This Article sketches a new vision for algorithmic accountability, with a more vibrant role for workers and for the public in determining how firms and government institutions work together.

John W. Bagby (Pennsylvania State University) and Kimberly Houser (University of North Texas) have posted “Artificial Intelligence: The Critical Infrastructures” on SSRN. Here is the abstract:

Artificial Intelligence (AI) innovation is most strongly impacted by AI Critical Infrastructures. These are the conditions, capacities, assets and inputs that create an environment conducive to the advancement of the AI technologies. Close inspection of AI’s generalized architecture reveals a supply chain that implies six AI critical infrastructures. There are at least seven necessary steps or processes contained in a generalized AI architecture. These steps are: (1) occurrences, events, facts or conditions transpire enabling the creation of potentially useful data, (2) these data are logged through capture and (increasingly computer and telecommunications enabled) initial storage, (3) such data are aggregated, often by numerous data repositories or AI operators, (4) human intelligence performs iterative analysis as derived from deployment of algorithms, (5) initial machine learning occurs, (6) near constant feedback loops are deployed by many AI applications that adapt the underlying model as new data is incorporated, and (7) based on insights resulting from AI, decision-making occurs, both automatically by computer or by human intervention,. Successful Machine Learning requires ample supply of the six broad AI critical infrastructures: (i) strategic insight/vision largely expressed as regional and/or national Industrial Policy, which is paramount in impacting all four other AI critical infrastructures, (ii) human intellect is needed to foster a deep-bench, from a competent AI Workforce, (iii) R&D Investment in AI, (iv) AI Hardware, both Computing Power and Connectivity (ICT), (v) bountiful and ever growing supply of Accessible Data, and (vi) market receptivity as sustainable demand for AI knowledge to monetize successful AI innovation. This article provides an initial foundation for a comparative of the three world economies (regions) seemingly best positioned to make substantial AI advancements. Predictably, significant differences among the political and cultural drivers in these three regions are likely to impact needed commitment to AI critical infrastructures: China (Asia) vs. the United States (North America) vs. European Union (EU). The harsh reality of AI innovation is that delays in commitment and deployment of AI critical infrastructures will relegate the losing region(s) to become, at best, a chronic AI customer rather than a major successful AI supplier.

Cary Coglianese (University of Pennsylvania Carey Law School) has posted “Moving Toward Personalized Law” (University of Chicago Law Review Online, Forthcoming) on SSRN. Here is the abstract:

Rules operate as a tool of governance by making generalizations, thereby cutting down on government officials’ need to make individual determinations. But because they are generalizations, rules can result in inefficient or perverse outcomes due to their over- and under-inclusiveness. With the aid of advances in machine-learning algorithms, however, it is becoming increasingly possible to imagine governments shifting away from a predominant reliance on general rules and instead moving toward increased reliance on precise individual determinations—or on “personalized law,” to use the term Omri Ben-Shahar and Ariel Porat use in the title of their 2021 book. Among the various technological, organizational, and political hurdles that stand in the way of a personalized system of law, I elaborate three obstacles that I refer to as the challenges of completeness, consensus, and currency. I then offer two solutions—custom and competence—that could bring about public acceptance to personalized law. Although I do not envision that these solutions can be complete ones, in the sense that they cannot prevent all problems with personalized law, a system of personalized law need not be perfect to be normatively appealing. All that personalized law must be is better than the imperfect rule-based system in place today.