Law After Singularity

4 October 2024

Imagine that tomorrow, OpenAI releases a model that surpasses human performance on every conceivable benchmark. It excels across every legal exam, and predicts legal outcomes in every jurisdiction with near-perfect accuracy. Its reasoning is so well-articulated that some advocate for its right to practise law—perhaps even to adjudicate cases. While such a prodigious model may not arrive tomorrow, the rapid evolution of Large Language Models (LLMs) compels us to ask: could AI ever truly replace a lawyer or a judge? In this talk, I explore the technical and intrinsic challenges that Legal AI must overcome before this point. Addressing the critical issue of the ‘black box’ nature inherent in neural models of law, I propose avenues for mitigating this opacity.

Join