He might be right on the big picture, since we're still far away from ever getting there, but it's just so hard to believe it based on what he writes, when so much of it is so, so, so, obviously incorrect.
Anyone could spend a day with Opus 4.5 and compare it to the frontier model of 12 months prior and immediately see it is a qualitative leap forward. The idea that LLM performance is the same now as it was a year ago is just insane.
Yeah, there are many many critical things to say about the current state of AI, the outrageous investments, the incessant hype and so on. But a lot of these staunch critics just disqualify themselves by throwing out statements that are clearly falsifiable, and if I notice that, why should I believe anything else they have to say.