Matthew Reid Krell (California State University, Northridge – Department of Business Law) has posted Inference: Legal, Scientific, And Artificial on SSRN. Here is the abstract:
One of the difficulties that law faces is the fact that it is, fundamentally, a humanistic discipline making judgments usually reserved for social sciences, about human behavior in the aggregate. Part of the difficulty is that law fundamentally does not make inferences in the ways that scientists, whether social or natural, recognize as valid. Proponents of artificial intelligence have argued that the gap between the two fields can be bridged with their large language models (LLMs). This Article explains the difference between the two methods of knowledge generation, and explains why the current architecture of large language models cannot perform inferences of either the legal or the scientific types, and why it is structurally incapable of being made to do so.