Abstract

We study the performance of ChatGPT interpreting prompts that require legal expertise to answer. Our inputs are very close adaptations from the "Contextual Canons" section of Antonin Scalia and Bryan Garner's Reading Law: The Interpretation of Legal Texts (Thomson West: 2012). We report our experiments and findings for the entire section (comprising 14 canons) of the book. We conclude that as a legal reasoner ChatGPT is exceptionally successful in taking the contextual canons into account.