GenAI as an International Lawyer: A Case Study with the Jessup International Law Moot Court

Abstract

This paper investigates the capacity of Generative Artificial Intelligence, specifically Large Language Models, to craft compelling international legal arguments. We tested the performance of two popular models, Gemini 2.0 and GPT4o, in the Jessup International Law Moot Court Competition, generating ten complete written memorials with minimal human intervention. With the organisers' blessing, these AI-generated memorials were anonymously added to the pool of submissions and evaluated by judges, who remained unaware of their origins, providing a unique benchmark against humanproduced work. Our results demonstrate that LLM-generated memorials consistently achieve average to superior scores, with some submissions receiving exceptional praise and near-perfect ratings. However, a detailed analysis of judges' qualitative feedback reveals persistent shortcomings of LLMs, notably factual inaccuracies, hallucinated citations, and superficial legal analysis. This study systematically identifies the current strengths and limitations of GenAI in legal argumentation, and critically informs best practices in prompt engineering, human-AI collaboration strategies, and emerging regulatory policies for legal education and practice.

Keywords: Generative AI, International Law, Jessup Moot Court, Moot Courts, Adjudication, Legal Argumentation, Large Language Models, Evaluation

JEL Classification: K10, K33, K41