I severely hope that people aren’t using LLM-AI to do reasoning tasks. I appreciate that I am likely wrong, but LLMs are neither the totality or the pinnacle of AI tech. I don’t think we are meaningfully closer to AGI than we were before LLMs blew up.