Legal AI’s Legibility Problem

Increasing reliance on language models for legal tasks have led to calls for more information about the likelihood and severity of potential errors. Realizing useful and robust transparency requires a