Susan Tanner (University of Louisville School of Law) has posted Prediction, Indeterminacy, And The Architecture Of Legal Meaning In The Age Of Generative AI (University of Louisville School of Law Legal Studies Research Paper Series, forthcoming) on SSRN. Here is the abstract:
Contemporary artificial intelligence systems can generate text that resembles legal analysis, which forces renewed attention to what counts as legal reasoning. This Article argues that in hard cases, legal reasoning is best understood as part legal reasoning and part public justification rather than mechanical deduction or mere prediction. We develop five interconnected claims: legal reasoning in contested contexts is typically enthymematic rather than syllogistic; the adversarial process is a social mechanism for identifying, testing, and stabilizing contested premises; AI performs best when premises and decision criteria have already been stabilized or when the legal domain is engineered for regularity; AI performs worst when the central dispute concerns the premises themselves and the task is justificatory; and this contrast clarifies the boundary between prediction and justification while highlighting how legal traditions differ in their prospects for technological integration. By analyzing how AI systems handle legal reasoning, we argue that the current focus on prediction in AI applications obscures the justificatory dimension of legal decision-making and risks undermining the legitimacy of legal institutions.
Recommended.
To receive a daily summary of posts from Legal Theory Blog by email, get a free subscription to Legal Theory Stack.
Lawrence Solum