Kind of got bored reading this article halfway through. Apologies to the author. But honestly, I'm going to remain in the "ChatGPT and Claude are just giant large probabilistic statistical machines", and very good ones at that -- They are (AFAIK) fundamentally incapable of new thought, new thinking or new innovation -- worse, they don't _actually_ think. Not like we (humans) do.

Far worse I think is the enormous cost and power to run these multi-thousand GPU clusters just to run these so-called LLM(s) models. Until you can get something approximating human-like intelligence and cognitive capability with the power consumption relative to that of the human brain, we're nowhere near close!

Don't get me wrong. I use Claude Code every day. It's great. But commodified intelligence is it not -- it is (and ChaptGPT) incredibly stupid.