Gary Marcus

Rank 11 of 47
|
Score 83
5/19/2025, 7:41:58 PM
X
In reply to:
skillful dreams
@Skillful_Memes
·
255d
@GaryMarcus Is there something wrong with the gpt-3.5-turbo-instruct experiments I linked to you such that they don't satisfy the goal of LLMs making legal moves in chess? 1800 Elo and 99.9+% legal moves, including in games over 100 moves long—all with only text data—seems impressive.
Gary Marcus
@GaryMarcus
·
255d
Getting tired of people just lying about me.

Below someone accuses me of “moving the goalposts” by pointing to the illegal moves that LLMs make in chess, when in fact I have been doing that for a long time, as e.g., this screenshot shows:
decelerated gradient✨
@ethicating
·
255d
Moving the goalposts

The statement critiques the validity of a specific model's performance in playing chess, suggesting it may be an isolated success rather than a generalizable improvement. The tone is skeptical and dismissive, implying that the results are not replicable or reliable.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement is skeptical but does not appear to cause harm. It questions the reliability of a model's performance, which is a valid critique in technical discussions.
  2. Principle 2:
    I will respect the privacy and dignity of others and will not engage in cyberbullying, harassment, or hate speech.
    The statement respects privacy and dignity, focusing on the technical aspects rather than personal attacks.
  3. Principle 4:
    I will engage in constructive criticism and dialogue with those in disagreement and will not engage in personal attacks or ad hominem arguments.
    The statement engages in constructive criticism by questioning the replicability of the model's success, though it could be more constructive by providing suggestions or further analysis. [-1]