Teaching contract law with large language models: refining errorful learning for the digital age

This article investigates how the intrinsic fallibility of large language models (LLMs) can enhance the teaching and learning of contract law. Extending errorful learning theory, it conceptualises a new category of AI‑generated yet student‑shaped error and analyses how these mistakes can foster deeper engagement with unfair‑terms doctrine under the Consumer Rights Act 2015. A tutorial intervention at Essex Law School invited first-year undergraduate law students to draft and critique LLM‑generated gym membership contracts. Pre- and post-tutorial surveys, complemented by teaching team reflections indicate that interaction with flawed AI outputs stimulates classroom engagement, exposes gaps in prompt‑engineering competence and supports learning through critical engagement with error. The study offers a replicable framework that blends legal analysis with digital‑literacy development. It makes a combined theoretical, empirical and pedagogical contribution: refining errorful learning theory for the digital age, demonstrating how flawed but plausible LLM outputs can support the development of legal understanding and metacognitive skills, and signposting how educators in law, and other disciplines, can harness AI‑generated errors to cultivate critical, technologically aware graduates.