Gary Marcus

Rank 12 of 47
|
Score 83
5/19/2025, 5:58:39 PM
X
In reply to:
Hermopolis Prime 馃嚦馃嚳
@HermopolisPrime
路
253d
@Skillful_Memes @GaryMarcus Looks like someone got even GPT 3.5 to play to a decent level, no doubt it could be done with o3 or Gemini 2.5
skillful dreams
@Skillful_Memes
路
253d
@GaryMarcus See this thread
Grant Slatton
@GrantSlatton
路
862d
The new GPT model, gpt-3.5-turbo-instruct, can play chess around 1800 Elo.

I had previously reported that GPT cannot play chess, but it appears this was just the RLHF'd chat models. The pure completion model succeeds.



See game & thoughts below:
Gary Marcus
@GaryMarcus
路
253d
Has anyone found an LLM that can reliably play chess, without making illegal moves?

The statement discusses the performance of AI models in playing chess, specifically mentioning their tendency to make illegal moves. It is a technical discussion about AI capabilities and limitations in a specific domain.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does not cause harm; it is a factual observation about AI performance.
  2. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement contributes to understanding AI limitations, promoting informed discussion.聽[+1]