Against AI Jurisprudence

Faculty Associate Dasha Pruss and coauthor Jessie Allen argue that LLMs provide politically illegitimate modes of juridical decisionmaking because they truncate the possibility of human discretion. They argue in a paper forthcoming in the Proceedings of the AAAI/ACM Conference on AI, Ethics and Society (AIES) that the chief problem with using LLMs to discern the 'ordinary meaning' of a law "is that it substitutes the aggregated subjective judgments of the surveyed humans, including whatever biases they harbor – or the LLM’s pre- diction of the collective accumulation of those subjective results – for the ultimate judicial determination of what the law is."