Artificial Intelligence and law

The law is a societally important and scientifically rich and challenging application domain of artificial intelligence. Much of my research in this field has been on computational models of legal argumentation. My early work was on the following topics.

• Logics for legal argumentation, in particular, my 1997 book (a revised and extended version of my 1993 PhD thesis). Subsequently I developed with Giovanni Sartor a logic for rule-based legal argumentation [Prakken & Sartor 1996].
• Dialogue game models of legal procedures; see, for instance, my AI & Law Journal article on A formal model of adjudication dialogues and an application of this model in a case study.
• Formal modelling of burdens of proof, with Giovanni Sartor, applying argumentation logics in the context of dialogue models of legal procedure; see e.g. our ICAIL-2007, JURIX-2011 and recently our ICAIL-2023-2011 papers.
• Legal reasoning about evidence. See, for example, two articles in Law, Probability & Risk (2004, 2014), my 2004 AI & Law Journal paper with Floris Bex, Chris Reed and Doug Walton on formalising evidential argument schemes, my ICAIL-2005 paper on accrual of arguments, an edited volume on Legal Evidence and Proof: Statistics, Stories, Logic, a co-edited special issue of Topics in Cognitive Science on Models of Rational Proof in Criminal Law, and two double PhD projects:
  • Making sense of evidence on software support for crime investigation;
  • Designing and Understanding Forensic Bayesian Networks with Arguments and Scenarios.
• AI & Law and autonomous systems. This research line concerns the problem of making autonomous systems behave in a legally and ethically responsible way. See my 2016 position paper on the general problem and my 2017 AI & Law Journal paper on making autonomous vehicles conform to traffic law.

My current work in AI and Law focuses on the following main themes:

• Modelling case-based legal reasoning and explanation. This work builds on my early work with Giovanni Sartor on formalizing factor-based reasoning with precedents [Prakken & Sartor 1998]. My current work includes a JLC 2015 paper with Adam Wyner, Trevor Bench-Capon and Katie Atkinson on formalising argumentation schemes for legal case-based reasoning in ASPIC+ and an AI & Law 2021 paper A formal analysis of some factor- and precedent-based accounts of precedential constraint.. I believe that this kind of work can help in explaining legal and other machine-learning applications through an analogy between legal precedents and training data; see e.g. [Prakken & Ratsma 2021], [van Woerkom et al. 2022] and [Peters et al. 2023].

• The computer judge. The recent successes in machine learning have led many in the legal world to believe that the computer judge is near. Some recent algorithms that predict outcomes of legal cases are often cited as evidence. I believe that these expectations are largely based on misconceptions of what such predictive models can do for the law. I have recently (co-)written two papers in Dutch journals about this: Prakken (NJB 2018, English translation) and Bex & Prakken (AA 2020, English translation). With Floris Bex I also have an ICAIL 2021 paper on the relevance of algorithmic decision predictors for judicial decision making, and a JURIX 2021 paper on whether predictive justice can improve the consistency of judicial decision-making. My NJB 2018 paper was cited by the Dutch minister of Justice in a letter to the Dutch parliament. I was interviewed about the 'robot judge' in the Dutch newspaper Trouw.

• Legal applications of large language models. Large language models are a revolutionary recent development in AI, with potentially many legal applications. On October 8, 2021 I commented on potential legal applications (in Dutch) at the annual meeting of the Dutch Association for Lawyers (NJV). Recently I published an overview paper On evaluating legal-reasoning capabilities of generative AI. On Sept. 9th, 2024 I gave my inaugural address (original Dutch version), (translated English version) at Utrecht University, in which I discussed the potentials and dangers of using large language models for legal reasoning tasks.

Associations, interest groups, resources, blogs, ...

• IAAIL - International Association for Artificial Intelligence and Law
• JURIX - Dutch Foundation for Legal Knowledge-based Systems
• computationallegalstudies.com, by Daniel Katz, Michael Bommarito and John Zelner.
• The North Sea Group. A research community on legal evidence and proof.

Workshops and Conferences

• 37th International Conference on Legal Knowledge and Information Systems (JURIX 2024)
  Brno (Czech Republic), 11-13 December 2024.

• 20th International Conference on Artificial Intelligence and Law (ICAIL 2025)
  Chicago (USA), 16-20 June 2025.

• 21st International Conference on Artificial Intelligence and Law (ICAIL 2026), Singapore, May or June 2026.

• International Conferences on Artificial Intelligence and Law (ICAIL).
• International Conferences on Legal Knowledge and Information Systems (JURIX).
• Summer Schools on Law and Logic, Fiesole, Forence (Italy).
• AI & Law Summer Schools. Fiesole, Florence (Italy).
• LEX Summerschools on managing legal sources. Florence/Ravenna (Italy).
• International Workshops on Juris-informatics (JURISIN): 2007-2018, 2019.
• International Conferences on Alternative Methods of Argumentation in Law, Brno, Czech republic: 2011, 2012, 2013.
• Workshop on Reasoning about Responsible Agency in AI
  Amsterdam (The Netherlands), 16-17 May, 2023.
• Symposium on Law and Artificial Intelligence: Not a World Apart, Utrecht (The Netherlands), 18 May 2022.
• 1st Workshop on Programming Languages and the Law (ProLaLa 2022), Philadelphia (USA), 16 January 2022.
• MIREL workshop on MIning and REasoning with Legal texts.
  Luxemburg, 17 September 2018.
• Conference on Recent developments in Artificial Intelligence: Law and Justice.
  Montreal (Canada), 27 October 2016.
• ECAI 2016 Workshop on AI for Justice.
  The Hague (The Netherlands), 30 August 2016.
• Workshop Responsible Intelligent Systems in Perspective; where Computer Science, Philosophy and Legal Theory meet
  Utrecht (The Netherlands), 18-19 April 2016.
• Zif Workshop on Models of Rational Proof in Criminal Law.
  Bielefeld (Germany), 28-30 September 2015.
• 2nd International Conference on Quantitative Aspects of Justice and Fairness (QAIJF 2011)
  Fiesole, Florence (Italy), 25-26 February 2011.
• Conference on Graphic and Visual Representations of Evidence and Inference in Legal Settings (conference brochure)
  New York, USA, 28-29 January 2007.
• Workshop `Dialectical Legal Argument: Formal and Informal Models'.
  Tilburg (The Netherlands), 12 December 1996.

Journals

• Artificial Intelligence and Law
• SCRIPTed: A Journal of Law, Technology & Society

Research projects

• Designing and Understanding Forensic Bayesian Networks with Arguments and Scenarios.
• Making sense of evidence.
  • Some press about this project:
    • The Dutch press about Susan van den Braak's PhD thesis on Sensemaking Software for Crime Analysis.
    • The Dutch press about Floris Bex's PhD thesis.
    • Floris Bex's weblog entry for a Cornell law School weblog.

Research Groups, Labs

• Dutch Police Lab AI.
• The Maastricht Law and Tech Lab..
• CODEX: Stanford Center for Computers and Law.

People