Gary Marcus

Rank 13 of 47

Score 84

@ai_in_check Not only that but I actually asked him to look into it :)

and his results are central to my next substack, coming Saturday!

6/26/2025, 3:47:42 PM

In reply to:

ai_in_check

@ai_in_check

225d

@GaryMarcus Gary, you'd love it

Mathieu Acher

@acherm

225d

The latest generation of reasoning LLMs perform worse at #Chess compared to previous models. o3 & o4‑mini vs weak Stockfish: illegal moves in 88% & 94% of 67 games. o3 breaks rules in 4 moves; both resigned while winning. Worse than GPT‑3.5‑turbo‑instruct (1750 Elo) Details ⬇️⤵️

Gary Marcus

@GaryMarcus

226d

What comes through to me over and over and over is that you never know whether any given prompt will return what you ask.

The statement is part of a lighthearted conversation about AI-generated images and their accuracy. It involves a playful exchange about the capabilities of AI in generating specific image prompts and a mention of upcoming content on a platform. The tone is casual and promotional, focusing on the quirks of AI rather than engaging in a substantive public issue.