Artificial Intelligence in Government and the Law

Artificial intelligence—technology able to perform tasks normally requiring human intelligence—is spreading across the United States. Artificial intelligence could drastically change the daily lives of people around the world. According to the National Science and Technology Council, research on artificial intelligence could “further our national priorities, including increased economic prosperity, improved educational opportunities and quality of life, and enhanced national and homeland security.”  

Both the Obama and Trump Administrations have recognized the potential power of artificial intelligence and the need for continued investment. As artificial intelligence continues to develop the technology has been used to support not just business, but also the public sector and the practice of law. 

Here is what scholars are saying about the use of artificial intelligence in government and the law:

Artificial Intelligence in Government 

• In a recent paper, Cary Coglianese of the University of Pennsylvania Law School and David Lehr of Yale Law School analyze the threshold question of whether algorithmic governance comports with the need for governmental transparency. They conclude that the requisite level of transparency in governance can be achieved by revealing a given algorithm’s purpose, design, and basic function, and assert that the use of machine-learning may ultimately both enhance accuracy and legitimacy in governance.
• Can artificial intelligence and related digital tools support judicial and administrative decision-making or administrative tasks? In a new research paper, Coglianese and Lavi Ben-Dor report on U.S. administrative agencies’ and the judiciary’s use of artificial intelligence. Although no courts or agencies have implemented algorithmic-based legal or policy decision making, several courts and agencies have launched protocols that use machine-learning tools in performing administrative tasks, according to the authors’ research. Coglianese and Ben-Dor conclude that the use of algorithmic tools by governmental entities shows future promise, if used responsibly.
• A Duke Law Journal article by Rory Van Loo of Boston University School of Law explains that regulators have advanced the use of artificial intelligence in regulatory tools assuming that such tools effectively monitor markets and require little government involvement after implementation. Van Loo cautions, however, that training “effective digital regulators” requires more resources devoted to oversight or research and development, perhaps through an interdisciplinary agency tasked with managing the technology.

Artificial Intelligence in the Law 

• In a recent paper in the Georgia State University Law Review, Harry Surden of University of Colorado Law School explores the capabilities and limitations of artificial intelligence—and the role it can and will play in the law. Surden argues that although artificial intelligence is useful for “highly structured areas where there are clear right or wrong answers,” it is not useful for “abstractions, understanding meaning, transferring knowledge from one activity to another, and handling completely unstructured or open-ended tasks.” For example, some attorneys use automated predictive coding systems to assist with document review, whereas some judges uses artificial intelligence software to predict recidivism in criminal defendants. Surden concludes that “it helps us have a realistic understanding of where artificial intelligence is likely to impact the practice and administration of law and, just as importantly, where it is not.”
• As the use of artificial intelligence grows, so too does the number of legal disputes about artificial intelligence. In a Brookings report, Melissa Whitney, of counsel at DLA Piper, argues that judges confronted with artificial intelligence-related litigation must be prepared to understand the relevant technical issues and how they intersect with the law. Whitney proposes that professional organizations such as state bar associations could lead efforts to familiarize attorneys with the technologies as well as relevant case law. She also suggests that judges seek the requisite training through processes that are as neutral as possible.