Professor Nikola Datzov Selected as Plenary Speaker at Stanford Law IP Conference

Professor Datzov shares new findings on AI and the future of judicial decision-making

Professor Nikola Datzov was recently selected from more than 125 presenters to speak during the closing plenary panel session at the 26th Annual Intellectual Property Scholars Conference (IPSC). IPSC is the largest academic IP conference in the United States and was hosted by Stanford Law School on August 6–7.

During the presentation, Professor Datzov shared the findings from a first-of-its-kind empirical study measuring the capability of AI to perform high-level judicial functions. In his forthcoming article, AIth Circuit Court of Appeals (co-authored with David Schwartz), he discusses some of his findings regarding AI’s capability to make recommendations and decisions for case outcomes.

After creating a novel dataset of more than 1,000 cases from two federal appellate courts and utilizing various frontier AI models, he was able to closely examine AI’s capabilities to perform various judicial functions, including categorizing cases, deciding case outcomes, and even writing judicial opinions.

By analyzing the outcomes of human and AI adjudicators for the same set of cases, his study revealed that AI models are highly capable for some cases and unreliable for others. Even more surprisingly, he found that AI’s confidence could reliably identify when AI is likely to be right or wrong. His research in this area has been widely recognized as providing pioneering contributions toward better understanding how AI is poised to impact our judicial systems in the near future.