Project

Artificial Intelligence and the Law

The Initiative on Artificial Intelligence and the Law (IAIL) is a Harvard Law School initiative based at the Berkman Klein Center. Directed by Oren Bar-Gill and Cass Sunstein, the initiative focuses on new challenges and opportunities for the law created by the rise of artificial intelligence.

While AI can make enforcement and adjudication more effective, potentially reduce discrimination, and make the drafting of contracts, briefs, laws, regulations, and court opinions faster and less costly, it also has serious implications for broad societal issues such as consumer protection; investor protection; false advertising; privacy; misinformation; and discrimination and civil rights.

The initiative will sponsor and promote new work on these topics by both faculty and students, hold conferences and symposia, and issue preliminary reports on emerging topics. A book by Bar-Gill and Sunstein on algorithms and consumer protection, developed at Harvard Law School, is slated to be one of the early products.

Advisory Board

The IAIL is overseen by an advisory board consisting of law school faculty including Chris Bavitz, Yochai Benkler, John Coates, Benjamin Eidelson, Noah Feldman, Lawrence Lessig, Martha Minow, Ruth Okediji, Holger Spamann, David Wilkins, Crystal Yang, and Jonathan Zittrain.

Mailing List

Subscribe to the IAIL mailing list to receive updates about the initiative.

Faculty Projects

Bar-Gill, Oren and Sunstein, Cass: Algorithmic Harm in Consumer Markets

Sellers and service providers are increasingly using machine learning algorithms to set prices for and target products to individual consumers. While the use of algorithms can benefit consumers in some cases, it might harm them in others. We identify the conditions for algorithmic harm in consumer markets and explore legal policy responses that could mitigate such harm.

Bavitz, Chris, Risk Assessment Tools Database

This project, at the Berkman Klein Center, tracks use of actuarial risk assessment tools in the criminal legal system in the United States (including in the context of pretrial release decisions).

Bavitz, Chris, Promises and Perils and of Artificial Intelligence for Creators

This project explores the promises and perils and of artificial intelligence (in particularly, generative AI tools) for creators, including impacts on the value of creative labor.

Lessig, Lawrence, AGI Circuit Breakers

How should we regulate to protect against misaligned AGI, and assure we retain control if capabilities exceed expectations.

Minow, Martha and Lévesque, Maroussia, Artificial Intelligence in the Judiciary

AI systems are increasingly used in adjudication and legal practice. We combine legal and technical expertise to map substantive, procedural and institutional implications and to suggest appropriate policy responses with a particular focus on uses of and policies about AI systems in courts.

Spamann, Holger, Automated Detection of Judicial Reasons

Co-author: Stammbach, Dominik

We automate the detection of reason types in judicial opinions by a combination of rules and the training of classifiers. This will allow the comparison of reasoning across time, space, and the judicial hierarchy. For now, we are working with English and German decisions; eventually, we want to expand to other jurisdictions including, of course, the U.S.

Spamann, Holger, Legal Text Deduplication / Simplification

Co-authors: Stuart Shieber, Elena Glassman, Corinna Coupette, Mirac Suzgu

Statutory language is often highly duplicative. This means the information density is low, but perhaps more importantly, it makes it hard for humans to comprehend the statute's text. We develop algorithms and GUIs that can shorten statutory text and improve comprehensibility with the help of humans-in-the-loop.

Yang, Crystal: Why Humans Override Algorithms

We are interested in understanding why humans override algorithmic recommendations and the welfare consequences of these overrides. We study patterns of overrides empirically and also plan to utilize surveys and qualitative interviews to better understand this phenomenon.

Zittrain, Jonathan, Adversarial Attacks on Medical Machine Learning

Co-authors: Samuel G. Finlayson, John D. Bowers, Joichi Ito, Andrew L. Beam, and Isaac S. Kohane

With public and academic attention increasingly focused on the role of machine learning in the health information economy, mastering these systems’ vulnerabilities becomes more critical. Even a small, carefully designed change in how inputs are presented to a system can completely alter its output, leading to wrong conclusions and dire consequences. These so-called adversarial attacks have expanded from a niche topic investigated by computer scientists to a conversation effecting the national healthcare system and with potential financial, legal, technical, and ethical implications. Far from discouraging continued innovation with medical machine learning, we call for experts across industries in pursuit of efficient, broadly available, and effective health care that machine learning will enable.

Zittrain, Jonathan, Intellectual Debt

Machine learning models the concept of “intellectual debt,” wherein we gain insights we can use immediately, and promise we’ll figure out the details later. Know now, understand tomorrow. Sometimes we pay off the debt quickly; sometimes, as with how aspirin lowers a fever, it takes a century; and sometimes we never pay it off at all. Loans can offer great leverage, be they money or ideas. Indebtedness also carries risks. For intellectual debt, these risks can be quite profound, both because we are borrowing as a society, and because new technologies of artificial intelligence – specifically, machine learning – are bringing the old model of drug discovery to a seemingly unlimited number of new areas of inquiry. Humanity’s intellectual credit line is undergoing an extraordinary, unasked-for bump up in its limit.