AgentCourt: Simulating Court with Adversarial Evolvable Lawyer Agents

TL;DR

AgentCourt introduces an adversarial evolutionary framework for simulating court proceedings with AI lawyer agents, significantly improving legal reasoning and reasoning agility through dynamic knowledge learning in a simulated environment.

Contribution

The paper presents AdvEvol, a novel adversarial evolutionary approach enabling dynamic knowledge learning and evolution of legal agents in simulated courtrooms, surpassing static models.

Findings

Achieved 12.1% performance improvement on CourtBench benchmark.

Evolved agents demonstrated enhanced legal reasoning and agility.

Professional lawyers confirmed the approach's effectiveness.

Abstract

Current research in LLM-based simulation systems lacks comprehensive solutions for modeling real-world court proceedings, while existing legal language models struggle with dynamic courtroom interactions. We present AgentCourt, a comprehensive legal simulation framework that addresses these challenges through adversarial evolution of LLM-based agents. Our AgentCourt introduces a new adversarial evolutionary approach for agents called AdvEvol, which performs dynamic knowledge learning and evolution through structured adversarial interactions in a simulated courtroom program, breaking the limitations of the traditional reliance on static knowledge bases or manual annotations. By simulating 1,000 civil cases, we construct an evolving knowledge base that enhances the agents' legal reasoning abilities. The evolved lawyer agents demonstrated outstanding performance on our newly introduced CourtBench benchmark, achieving a 12.1% improvement in performance compared to the original lawyer agents. Evaluations by professional lawyers confirm the effectiveness of our approach across three critical dimensions: cognitive agility, professional knowledge, and logical rigor. Beyond outperforming specialized legal models in interactive reasoning tasks, our findings emphasize the importance of adversarial learning in legal AI and suggest promising directions for extending simulation-based legal reasoning to broader judicial and regulatory contexts. The project's code is available at: https://github.com/relic-yuexi/AgentCourt