pull down to refresh

Silicon Formalism: Rules, Standards, and Judge AISilicon Formalism: Rules, Standards, and Judge AI

AbstractAbstract

This paper replicates a judicial experiment originally conducted on 61 U.S. federal judges, this time with GPT-5 as the decision-maker. The experiment involves a choice-of-law dispute in a hypothetical automobile accident case, where three variables are manipulated: (1) whether the applicable doctrine is a rule or a standard, (2) whether the plaintiff or defendant is portrayed more sympathetically, and (3) the location of the accident, which affects the legal outcome under different states' choice-of-law rules. Consistent with our prior work, we find that the LLM adheres to the legally correct outcome significantly more often than human judges. In fact, the LLM makes no errors at all.

https://download.ssrn.com/2026/1/30/6155012.pdf?response-content-disposition=inline&X-Amz-Security-Token=IQoJb3JpZ2luX2VjEGoaCXVzLWVhc3QtMSJIMEYCIQCUKpchvrUkvSavhzGIeu5HZSxFK18GWPDY%2Fa8mGSRbLQIhALj%2BipyCCHh8%2Bq2QudVJNpPhy7Urff19Ir38xbk5xyADKr0FCDIQBBoMMzA4NDc1MzAxMjU3IgyXVa5lf9xenlbs9%2B0qmgWE9h5de0RCzWkZNuv7qRUduy8BVuRs3s3kO%2FOX3K0u%2FQpRNoZWpoSCG7jjBzhVLu5Y756HLmzM6PqDlE2KQLKjxmFAooC3BA7x%2BrdXOGuHZfPzxEg8W1%2F08TOgioiNL25hHhNMnTVh5WTMJQ2198aBwHMBpuDJD1fTfAqnH2AGyqvLL6l93Uz2kb1Etlkvs3Xw%2FTP30Lba6os%2F2lOJasAf620VOTB5lxgW1JI39tw%2FITPThcCS3q8dUhrVFxF8pvR2l17PsV7iU%2BXw0FIjJrex2wniajFpjhbQRQvFPTuopxGgW8V0rmRMw2SkKUtXbKOCGYtGlqnuZ7FTGXQbqQSFYfOaUspaFKX%2Fq6hrN%2Fo5eokMqHaqkiP2Mh0pQR4J7GwfDrwdw8ZXhq6H%2Fg1LewRbcBBx6AB0NgqjzvgR18THxhi1F%2Bb51NH2d38zaOdz3W0T4M%2FUJBIbE%2BSCy01iz1jacbTNt%2FbLCO7HykQSmdVIqeLy2SLA0L8TuNIn6mJ3SWQZFArbcSwFcrCskVC9mxMSgNvhCaMxpBPSKkuIIQJWX3iiOIPz6H21TIbhV7BHGY4OfAT2PntCrVCrq8XcbnvxH2D1Kr730q2IDGtQHNBZBThEEQVsE6TTG4IhSdmgCJlc84HxMewVT5FYUQU%2BnoaXr7NeqbrYinnp7zHWBJZy4%2FjwvKUNb5LOpVf6t1xqXdKHExaKMjdhFV%2BnkKyiHNuqWN7vo58qWn9skyRjvq3Rt0z24ghkrVvinncunV%2FJ2h0J9h7Q3hWJq4sJJlthfYkrsDvu3FDF2kBPHTQRc9xcDVRr18%2FwaLyCAyKu8MYVaMIX7k08NLU0FLH6ZfqC86lBLSO78PU6L8e%2FHfNxh9YRnQ905SmRYLpXiIIw1cLLzAY6sAGuclUz2%2BMyZyv217c%2BAIw%2FrFfCHCbB6V%2BN5kJn1pShdiLX24Oy%2FAmX2%2B4U1Sx5vuws2kNii92lFWLF3vDWQFCUz0hM%2BpejXsdM3jjeRh0D1RoeQ0ALIVu8e22%2BupzeqArWLBxUim%2BelXquxH%2Fij2N%2FgjzEtYCUG1dgP8l1dTKWyxNashabG38wo%2FH3A2ZUDj%2Fm0qxpsJcLRcMYadBXyK1m5Y0Tg%2BuI90hjp6MDdQD58w%3D%3D&X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Date=20260216T104508Z&X-Amz-SignedHeaders=host&X-Amz-Expires=300&X-Amz-Credential=ASIAUPUUPRWE5OSJFL5W%2F20260216%2Fus-east-1%2Fs3%2Faws4_request&X-Amz-Signature=7951344c933d4a4efdce3d70f2cd46665f2f177069de39058082a28c4f01caf5&abstractId=6155012

No surprise here. Federal judges decide on an outcome they want, then grab their recent law school graduate clerks to justify their opinion in a written decision. Federal judges are political animals.
Some are smart. Some are not so smart.

reply

Confirming what we all already knew?

I'm willing to give the AI overlords a try. It can't be worse than being ruled by pedophiles

reply

AI (LLMs) are basically just people with a bit of fine-tuning!

reply