started · updated
UNC Law AI justices mirror US Supreme Court ruling outcome
During a mock argument at the University of North Carolina School of Law, three ChatGPT-based AI justices reached the same legal conclusion as the US Supreme Court regarding a 2025 immigration case, Monsalvo Velázquez v. Bondi.
The case centered on whether a 60-day voluntary-departure deadline falling on a Saturday should extend to the following Monday. While the actual Supreme Court decision resulted in a 5-4 split, the AI justices reached a 2-1 decision. The AI models were programmed to follow different interpretive frameworks: a textualist adhering to literal meaning, a pragmatist focusing on practical results, and a progressive considering modern social values.
Developed by data librarian Rolando Rodriguez, the AI judges operated offline using anonymized legal records to maintain neutrality. While the machines were able to reason and pose questions, participants noted that the AI lacked the natural back-and-forth characteristic of professional oral arguments. Nicole Downing, assistant dean for AI at UNC Law, stated the exercise aimed to test AI's reasoning capabilities rather than to replace human judges.
Entities
ChatGPT · Nicole Downing · Rolando Rodríguez · US Supreme Court · University of North Carolina School of Law