Towards Autonomous Legal Intelligence: Case-Based Learning, Legal Rule Induction and Agentic Reasoning

PhD Thesis Proposal Defence


Title: "Towards Autonomous Legal Intelligence: Case-Based Learning, Legal 
Rule Induction and Agentic Reasoning"

by

Mr. Wei FAN


Abstract:

A legal conclusion is inseparable from the conditions that justify it: who 
acts, under what circumstances, and which law applies at the relevant time. 
Yet language models often reproduce legal outcomes without reliably capturing 
these conditions. This thesis investigates how the conditions of legal 
applicability can be learned from cases, generalized into rules, and checked 
through autonomous research. Cases synthesized from privacy laws expose the 
contextual distinctions needed for learning, while induction across analogous 
precedents reveals principles that extend beyond individual decisions. 
Autonomous research adds a further challenge: deciding what evidence to seek 
and which legal authorities govern the case. Training agents through 
reinforcement learning addresses both research planning and the temporal 
consistency of legal search. The proposed research will connect these 
capabilities by investigating how induced rules can guide evidence gathering 
and how new evidence can refine the application of those rules. The aim is a 
foundation for autonomous legal intelligence whose conclusions remain 
grounded as the available facts and governing law change.


Date:                   Wednesday, 21 October 2026

Time:                   4:30pm - 6:30pm

Venue:                  Room 5504
                        Lifts 25/26

Committee Members:      Dr. Yangqiu Song (Supervisor)
                        Dr. Chaojian Li (Chairperson)
                        Prof. Nevin Zhang