Can generative artificial intelligence identify case-law rendered vulnerable by the Revocation Act 2023?

Abstract Parts of domestic UK case-law rely in one way or another on European Union (EU) law. The authority of such precedent has potentially been greatly diminished by the Retained EU Law (Revocation and Reform) Act 2023. This study investigates whether the Large Language Model (LLM) ChatGPT-4o is capable of reliably distinguishing judgments where the ratio turns on direct effect or conforming interpretation from those where it does not. It finds that, given a specific prompt, ChatGPT masters the task with a reliability of more than 90 per cent. This result may be an indicator that LLMs can, more broadly, play a productive role in the legal field for the purpose of identifying risks and weaknesses in legal reasoning.