One can argue that the art of medicine lies in phronesis. Aristotle differentiated this practical wisdom from episteme (scientific knowledge) or techne (technical skill). It is not just things to know or things to do, but the combination of perception, judgment, and application: the salience to grok a situation and its pertinent features and then apply yourself to deciding your course of action.
Maybe we’ll eventually see AI, as Anthropic CEO Dario Amodei argues, as the world’s best at everything. Not an average white-collar worker, but the best. Before that, AI is probably more likely to be “good enough” for a bunch of stuff. It will make it easy for us to fall back to procedural decision-making, which will get more robust and even more defensible. A reversion to the world’s most boring mean.
But forcing problems into an algorithm does not resolve intrinsic uncertainty; it just helps absolve us of individual responsibility and associated guilt. It removes judgment, which can be “helpful” because decision-making is scary, even if no individual patient is actually the average. We work in consultant-filled hospitals with hospitalist orchestrators partly as an elaborate risk-transfer game. But the ultimate risk transfer is “best practice.” The less human our practice, the easier it is to replicate mechanistically.
If there is something durably unsatisfying about AI-for-everything, it will lie in systemically repudiating phronesis.