analyzed the limits of large language models in a Thursday post on X. He wrote that the field produces "overly compliant helpers" rather than revolutionaries. Right now, AI isn't creating new ...