Gary Marcus

Rank 11 of 47
|
Score 83
12/31/2024, 4:37:07 PM
X
In reply to:
Alexandre Chiavegatto Filho
@SaudenoBR
·
395d
@GaryMarcus @TomerUllman A major one is also that you will always try to find a nuance on your previous takes to convince people you were never wrong.
Gary Marcus
@GaryMarcus
·
395d
@SaudenoBR Incorrect. Compositionality and relations were among the central issues I laid out in Deep Learning is Hitting a Wall and they are s not solved. see https://garymarcus.substack.com/…

See e.g., https://garymarcus.substack.com/…, and recent work on relations by @TomerUllman cited there.
Alexandre Chiavegatto Filho
@SaudenoBR
·
395d
@GaryMarcus Another issue most have with your takes is that you keep moving the goal post regarding which are your major issues with LLMs, as they keep solving them.
Gary Marcus
@GaryMarcus
·
1256d
More evidence that deep learning is, for all the progress, still stuck when it comes to relations and compositionality.
Gary Marcus
@GaryMarcus
·
395d
@SaudenoBR In sum, you said “quite a few”, but gave only one, and that was based on an erroneous understanding of the original paper.
Gary Marcus
@GaryMarcus
·
395d
@SaudenoBR i didn’t say it would make no further progress, I said it would get stuck on certain problems, which still remain, and that scaling would was not a physical law and would not hold indefinitely or solve those specific problems; please reread the actual paper carefully and my line…
Alexandre Chiavegatto Filho
@SaudenoBR
·
395d
@GaryMarcus There are quite a few, the most famous is probably when you said deep learning was starting to hit a wall in 2022 (before the release of ChatGPT)

The statement engages in a discussion about the interpretation and critique of previous arguments related to deep learning. It addresses the issue of misrepresentation and the challenge of maintaining nuance in public discourse.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement attempts to clarify previous positions and address misunderstandings, aligning with the principle of striving to do no harm with words. [+1]
  2. Principle 2:
    I will respect the privacy and dignity of others and will not engage in cyberbullying, harassment, or hate speech.
    It respects the privacy and dignity of others by focusing on the content of the argument rather than personal attacks. [+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 in constructive criticism by addressing the issue of misrepresentation and defending the original argument without resorting to personal attacks. [+1]