Title: Hybrid decision making using knowledge and data
Abstract: The field of artificial intelligence (AI) is rapidly developing, with the public availability of large language models a recent significant breakthrough. The developments also affect the uses of AI
for decision making. Concerning decision making in the field of law, we read in the press that an AI system can now pass the bar exam, and also that an experienced lawyer submitted a computer-generated document to court---where
the judge discovered that the text was legal nonsense.
This talk addresses recent developments in AI and their implications for decision making. Specific attention is paid to the connections between logic-based AI in which the emphasis is on knowledge representation and reasoning, and data-driven AI aimed at machine learning. It is discussed how the combination of knowledge and data inspires hybrid approaches to AI-supported decision making. Specific attention is paid to the idea of argumentation systems, in which hypotheses are generated and critically discussed.
Bio: Bart Verheij holds the chair of AI and argumentation as full professor at the University of Groningen, and is the head of the department of AI in the Bernoulli Institute for Mathematics, Computer Science and Artificial Intelligence. He is the president of COMMA, past president of IAAIL (2018-2019) and vice president of JURIX. He is also co-editor-in-chief of the journal Argument and Computation, and section editor of the journal Artificial Intelligence and Law. Inspired by the domain of law, he uses an argumentation perspective to investigate the connections between knowledge, data and reasoning, as a contribution to responsible AI. He has significant achievements in the logic of argumentation, including extended semantics for abstract argumentation, case models for evidential reasoning and ethical system design, and modeling of reasoning with arguments, rules and cases in the law. He also works on the design and logical underpinning of argumentation software, and developed several systems of automated argument assistance.
Title: Symbolic and Sub-symbolic AI for Legislation
Abstract: The recent LLM and generative AI have been creating opportunities and risks that the AI and Law community should address with respect to the theory of law, the Rule of Law, and the digital transformation
of society. AI techniques applied in the Parliamentary environment require additional parameters for guaranteeing democratic principles, transparency, and explicability. A hybrid AI (i.g., neuro-symbolic AI), using symbolic
and non-symbolic approaches, is necessary for mitigating the risks and in the meantime fostering the potentialities of the technical evolution in the legislative domain. The talk presents the ERC HyperModeLex in the foundation,
research questions, and the main theoretical issues in the tension between symbolic and non-symbolic AI in the legal domain.
Bio: Monica Palmirani is a full professor of Computer Science and Law at the School of Law, University of Bologna. She is the vice president of IAAIL, member of the editorial board of Artificial Intelligence and Law, and director of the LAST-JD program. Her principal field of research is in legislative and legal informatics, in particular XML techniques for modeling legal documents both in structure and legal knowledge aspect, as well as legal drafting techniques supported by ICT. She has made prominent contributions to promoting the use of Legal XML in public agencies. In this regard, she has managed or coordinated many renown European initiatives, including COMPULAW, INTERLEX, Estrella and SEAL. She is also the co-chair of LegalRuleML and LegalDocML OASIS TCs, both of which are leading projects in the field of Legal XML. She was awarded the OASIS Distinguished Contributors in 2015 for her outstanding achievements in advancing open standards.