Philosophical Ideas

The deepest fault line in AI philosophy maps exactly onto a 2,400-year-old argument the Dreyfus brothers traced back to Plato and Aristotle. Plato bet that wisdom could be spelled out in explicit principles; Aristotle said no, underneath all your rules sits a judgment you can't formalize — ask a wise old man, not a philosopher. Dennett, watching the same debate replay in the 1980s between East Pole formalists (Fodor, Chomsky) and West Coast holists (Dreyfus, Searle), noticed something the partisans missed: AI researchers weren't replacing philosophy, they were doing it under tighter constraints, forced to actually build the schemata that Kant and Leibniz could only gesture at. What's strange is that large language models have now done something neither side predicted — they achieve Aristotelian-looking practical judgment through essentially Platonic means, by ingesting enough explicit pattern that something resembling the wise old man's intuition falls out. The seesaw didn't resolve. It fused.