Gary Marcus

Rank 13 of 47
|
Score 84
6/27/2025, 5:20:48 PM
X
In reply to:
Marcus Arvan
@marcusarvan
·
228d
@GaryMarcus @_vincentpaul_ @eshear {cough} https://philpapers.org/rec/ARVIA…
Gary Marcus
@GaryMarcus
·
228d
@_vincentpaul_ @eshear it’s like discovering that your knife can suddenly jump off the table.

if you read the paper, the outcomes are very very far from the training. even the authors didn’t expect the specific result.
Vincent Paul
@_vincentpaul_
·
228d
@GaryMarcus @eshear Curious about your take. The study suggests narrow fine-tuning on malicious tasks can broadly misalign LLMs. Interesting, but a "No duh" moment for me? Like blaming a knife because it can cut. Powerful tools inherently carry risks; thoughtful use and design remain key.
Gary Marcus
@GaryMarcus
·
229d
Vile.

ChatGPT should be pulled from the market til they fix this.

The statement engages in a discussion about the alignment and safety of AI models, specifically addressing concerns about their potential misuse and the implications of their outputs. The tone is analytical and open to further exploration, indicating a willingness to engage in deeper analysis.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement acknowledges the potential harm of AI models but does not directly address how to mitigate it, showing partial alignment with the principle of doing no harm. [+1]
  2. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement promotes understanding by recognizing the complexity of the issue and expressing an interest in further exploration, aligning with the principle of promoting empathy and understanding. [+1]
  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 constructively with the previous argument, acknowledging its directionality without resorting to personal attacks, aligning with the principle of constructive criticism. [+1]