Michael Showalter (Independent) has posted The Honorable Robot: AI Adjudication and the Judicial Virtues (86 Ohio State Law Journal Online 1 (2025)) (17 pages) on SSRN. Here is the abstract:
In a 2003 article titled Virtue Jurisprudence, Professor Lawrence Solum develops a “virtue-centered theory of judging.” In this article I examine the judicial virtues and vices Solum identifies and conclude that properly trained AI adjudicators may become more virtuous on these metrics than their human counterparts. For the most part, my conclusion seems obvious—on intelligence, temperament, and impartiality, for example, AI has clear advantages.
The real question is whether there is something uniquely human about judging that machines cannot replicate. The only virtue Solum lists that seems like a plausible candidate is what he calls “judicial wisdom”—a form of phronesis, or “sound practical judgment.” It’s not entirely clear what this means or how it is conducted. If judicial wisdom means the capacity to identify the legally salient facts and faithfully apply the law, then it doesn’t seem to be a distinct virtue. If judicial wisdom means something more—something moral or extralegal—then it is highly controversial whether it is a virtue at all.
As this highlights, machine adjudication implicates deeply theoretical and longstanding debates over the nature of judging. Those who believe that judging sometimes requires the judge to answer moral questions—whether because they believe law sometimes “runs out,” or the law sometimes grants policymaking discretion, or consequentialism requires it—might be more likely to resist machine adjudicators that might not share humans’ moral vision. Those who believe that feelings and emotions should influence judicial decisions, similarly, will more likely favor adjudicators who have those things. On the other hand, those who believe that judges should not allow personal views, feelings, or emotions to influence their decisions will be more likely to favor machine adjudication. Those who view judging as a wholly impartial, blindfolded task will likely favor machines that can be trusted to keep the blindfold on.
Posted at 6:00 AM