Gary Marcus

Rank 11 of 47
|
Score 83
5/19/2025, 7:44:01 PM
X
In reply to:
Andrew Prock
@andrewprock
·
255d
@PurensPhD @littmath @GaryMarcus @colin_fraser @MelMitchell1 @rao2z I suspect this really comes down to undefined terms over a subjective domain. Arguing that models are (or are not) intelligent is kind of like arguing they are (or are not) loving.
Kristopher Purens
@PurensPhD
·
256d
@littmath @GaryMarcus @colin_fraser @MelMitchell1 @rao2z they’re not ‘hard for frontier models’ they’re just hard for intelligence as we know it.

The weirdest cope to come out of LLMs is very intelligent people finding some thing they are better than the models at and then declaring the models not intelligent.
Daniel Litt
@littmath
·
257d
@GaryMarcus @colin_fraser @MelMitchell1 @rao2z They’re hard proof questions, which are indeed difficult for frontier models. There’s nothing to explain.
Gary Marcus
@GaryMarcus
·
257d
@littmath @colin_fraser @MelMitchell1 @rao2z So how do you explain the much worse results on USAMO linked above?
Daniel Litt
@littmath
·
257d
@GaryMarcus @colin_fraser @MelMitchell1 @rao2z I also ask lots of other problems, including subtle alterations of the FrontierMath problems and totally unrelated ones. IMO this is wrong. (I also think the idea that they’re training on FrontierMath is very unlikely.)
Gary Marcus
@GaryMarcus
·
257d
@littmath @colin_fraser @MelMitchell1 @rao2z But OpenAI has very likely been augmenting around those problems - which they have privileged access to - and no system did well on the USAMO when data leakage was controlled.

(Also I am betting one could subtly alter those problems and get wild results.)

The statement is dismissive and sarcastic, questioning the intelligence and sincerity of others in the conversation. It does not engage constructively with the ongoing debate about the intelligence of models and instead resorts to personal attacks.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does harm by belittling others, violating the principle of doing no harm. [-2]
  2. Principle 2:
    I will respect the privacy and dignity of others and will not engage in cyberbullying, harassment, or hate speech.
    It disrespects the dignity of others by implying their intelligence and sincerity are lacking, bordering on harassment. [-2]
  3. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement does not promote understanding, empathy, or compassion, as it dismisses others' views without constructive engagement. [-2]
  4. 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.
    It fails to engage in constructive criticism or dialogue, opting for personal attacks instead of addressing the argument. [-2]