<p>Scholarship and policy on the use of artificial intelligence in legal judgment continues to be shaped by a consensus formed at an earlier stage in AI’s development. AI cannot be sensitive to context, reason normatively, or be responsive to novel facts. Recent experiments challenge these assumptions by showing that language models prompted to decide a case based on party materials uploaded to the model can render a reasoned decision comparable in quality to that of a human. This new capability of AI casts in a new light philosophical reservations about algorithmic judgment based on ideas about normativity, Aristotle’s concept of <i>phronesis</i> (practical wisdom), and Hannah Arendt’s theories common sense and reflective thought. Revisiting these arguments in the context of AI’s new capabilities illuminates how differences between human and algorithmic judgment are evolving and supports a pragmatic alternative basis for accepting the language model as judge in some cases.</p>