the most important AI startups won't be the ones that replace humans... They'll be the ones that help humans figure out what to do next what will humans do next? I think we get a clue when it comes to the discussion around "taste" on one end of the spectrum and verifiability on the other AI is great at verifiable problems. Did the code compile, is the math right. Where it struggles: "was that email rude?" "does this tweet hit hard?" the gap between objectively verifiable and subjectively verifiable is where the most interesting companies will be built, because it's where humans can team up productively with AI