Gary Marcus

Rank 12 of 47
|
Score 83
1/1/2025, 1:15:29 PM
X
In reply to:
Surya Ganguli
@SuryaGanguli
·
396d
@GaryMarcus They don’t always memorize.
Surya Ganguli
@SuryaGanguli
·
396d
Our new paper! "Analytic theory of creativity in convolutional diffusion models" lead expertly by @MasonKamb https://arxiv.org/abs/2412.20292
Our closed-form theory needs no training, is mechanistically interpretable & accurately predicts diffusion model outputs with high median r^2~0.9
Gary Marcus
@GaryMarcus
·
396d
And to think that a Nobel Prize winner told me Generative AI systems don’t memorize, because “they just store weights” 🤦‍♂️
Ed Newton-Rex
@ednewtonrex
·
396d
Generative AI models memorize their training data, exhibit 31692

The statement engages in a public discourse about the capabilities and limitations of generative AI models, specifically addressing the debate on whether these models memorize their training data. The tone is critical and somewhat dismissive, particularly towards Geoffrey Hinton's argument.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does not directly cause harm but uses a dismissive tone, which could be seen as undermining constructive dialogue. [-1]
  2. Principle 2:
    I will respect the privacy and dignity of others and will not engage in cyberbullying, harassment, or hate speech.
    The statement does not engage in cyberbullying or hate speech, but the condescending tone could be seen as disrespectful.
  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 does not engage in constructive criticism and instead uses a personal attack on Hinton's argument, which could hinder productive dialogue. [-1]