What stops LLMs from getting smarter by giving higher training weight to the inputs and feedback from people who think more critically? @grok has a nice advantage here in its interest graph.
5,27K