But the trend is clear: answers to even very hard questions are becoming cheaper and cheaper and cheaper. Which means the ability to ask them is getting more and more and more valuable.
What a modern LLM does during training is, essentially, very very quickly skim the textbook, the words just flying by , not spending much brain power on it.
Rather, when you or I read that math textbook, we read a couple pages slowly; then have an internal monologue about the material in our heads and talk about it with a few study-buddies; read
Actually, fuck answers, the more I think about it. The best questions don’t have answers. The point of the question isn’t to find an answer. The best questions are organizing principles, magnets, ways of seeing.
I’m a fan of companies that operate like studios with the awareness that they’ll create a series of seasonal hits and try to capture value along the way. The assumption of seasonality could be wrong (and one or more of these hits endures much longer), but that ends up being a good problem to have.