About this Series
As part of this compilation of educational materials for judges, AAAS offers an innovative set of three podcasts, focused on various ways in which modern technologies, usually AI or ML based, are affecting the legal field, including decision-making by courts, administrative bodies, and law firms. In each podcast, legal and AI experts discuss a variety of matters related to the podcast’s particular subject area, with the aim of familiarizing judges with key concepts and questions that should be borne in mind.
The first podcast looks at “risk scores” – what might be in the mix of factors that, in turn, are assessed to produce a “score”; and how such risk scores may be used, or misapplied, at different stages of legal proceedings (e.g., pretrial release, sentencing, etc.), particularly in criminal law. It emerges that a lack of standards as regards what goes into the “secret sauce” of various risk scores, and the possibility that scoring tools designed for one purpose might be used in unanticipated ways, mean that further thoughtful dialogue between the worlds of technology and law is imperative.
The second podcast brings on board special guests – from the private sector – who provide a survey of how AI can be used, notably by companies and law firms, to identify potential issues or even legal strategies, during, or in anticipation of, litigation. Among other things, AI-backed “solutions” can help companies and firms rapidly review mountains of data. Indeed, failure of counsel to use AI tools in the course of E-discovery, nowadays, might constitute a lack of professionalism. But, as the experts note, AI requires on-going human oversight, even as it transforms the legal field and even, potentially, courtroom procedures.
The third podcast asks, in effect, the question: Will AI ever replace humans, notably judges, as decision-makers? The answer is, however, not simple: Whereas AI is already being used in certain circumstances (e.g., by administrative bodies or, sometimes, in cases involving limited amounts of money), our experts recognize that complex cases, or cases involving high stakes, are unlikely to be left to automated decision-making in the foreseeable future. Nevertheless, the use of AI tools at various stages of legal work and court proceedings will only increase with time. This might take the form, one day, of AI helping parties or judges evaluate the credibility of witnesses, or of improving justice through a process of “conjoined” decision-making whereby machines and humans play to their strengths to reduce bias, improve efficiency, and enhance the accuracy of findings. Further dialogue and understanding between the technology and legal spheres constitute, again, prerequisites for using AI in ways that might enhance justice.
The production of this podcast series and other materials in the Artificial Intelligence and the Courts: Materials for Judges series was funded by the United States National Institute of Standards and Technology (NIST) through Award 60NANB21D031.
Disclaimer: The opinions, findings, and conclusions or recommendations expressed in these recordings do not necessarily reflect the views of the AAAS Board of Directors, its Council and membership, or NIST.
List of Episodes
Episode 1: AI and Risk Scores
Description: To what extent are courts or administrative bodies using systems that purport to provide insights into what a person will do, or should receive – e.g., in terms of imprisonment or benefits? What goes into “risk scores” that might be designed for use at differing stages of judicial or administrative proceedings? Might risk scores contain questionable factors, or be designed for one use yet – due to “function creep” – be used inappropriately? Our experts consider such questions in this section of educational materials for judges.
Episode 2: AI in the Legal Field - Commercial and Unexpected Uses
Description: This podcast focuses on several uses of AI in the legal field of particular importance – in the first instance – to law firms and companies. For example, companies and law firms use AI-backed tools, or “solutions,” to shift through mountains of data to fulfill modern expectations for “E-discovery,” or to protect intellectual property through monitoring numerous websites simultaneously, or to identify other issues that only can emerge from AI’s ability to find patterns in large volumes of data, very rapidly. Yet, as our special guests and experts acknowledge, the process of developing such AI tools, and of determining the ultimate utilization of AI-generated insights, require human involvement or oversight. That is, commercial applications of AI do not necessarily prove anything; rather, AI-powered analytics can flag important patterns or issues, which might constitute a first step towards bringing a matter before a court. Similarly, AI’s ability to help assess work and information flows might, someday, be leveraged by more courts to support their ceaseless efforts to ensure the rule of law.
Episode 3: AI, Decision-Making, and the Role of Judges
Description: The increasing use of AI – at various stages of judicial and administrative proceedings – may provide for more rapid, and less biased, decision-making but, when and where should humans remain “in,” or “on,” the “loop” of decision-making? Should factors such as the stage of proceedings, or the amount of money involved, or the severity of risk to individuals in case of AI error (e.g., imprisonment), or public expectations of the legal system, be taken into account? Indeed, what questions are AI “solutions” trying to answer? Do inherently biased humans (jurors, administrators, or judges), remain vital to ensure justice even while leveraging AI to enhance efficiency?