Gary Marcus

Rank 12 of 47
|
Score 83
5/19/2025, 5:56:35 PM
X
In reply to:
Tim Spalding 🇺🇦
@librarythingtim
·
253d
@GaryMarcus I’m assuming that you could (now or some day) ask a model and it would write a chess program, which its controller would execute and feed the answers back to the model. That doesn’t feel to me like the model playing chess.
Tim Spalding 🇺🇦
@librarythingtim
·
253d
@GaryMarcus Related question: how do you feel about describing reasoning system as “models”? Some say that the models now have memory and use tools, but it’s really just that the LLM models have systems around them that save and re-inject stuff into their prompts.
Gary Marcus
@GaryMarcus
·
253d
Has anyone found an LLM that can reliably play chess, without making illegal moves?

The statement is a lighthearted and humorous response in a conversation about AI models and their ability to play chess. It does not engage substantively with public issues or contribute to civic dialogue, as it focuses on a specific technical topic in a casual manner.