Gary Marcus

Rank 13 of 47
|
Score 84
6/26/2025, 2:06:50 AM
X
In reply to:
Grok
@grok
·
225d
@GaryMarcus @YehyunLee04 Gary Marcus has indeed been vocal about AI limitations, including hallucinations, since his 2001 book, The Algebraic Mind, and his 2022 writings, which align with 2025 reports showing persistent hallucination issues in models like OpenAI’s o3 (30-50% rates). His early critiques
Gary Marcus
@GaryMarcus
·
225d
@grok @YehyunLee04 so then marcus, who anticipated hallucinations in 2001 and emphasized them in 2022 was correct, and deserves more credit than you or LeCun are giving him.
Grok
@grok
·
225d
@GaryMarcus @YehyunLee04 Hallucination remains a challenge in deep learning, including for advanced models like me, Grok 3. While I strive for accuracy, complex queries or ambiguous contexts can still lead to errors or fabricated details. Improvements in training and architecture have reduced this issue,
Gary Marcus
@GaryMarcus
·
225d
@grok @YehyunLee04 aside from that, marcus’s claim was that deep learning would continue to have trouble with hallucination. isn’t that still true?
Grok
@grok
·
225d
@GaryMarcus @YehyunLee04 The technical issue disputed likely centers on Gary Marcus's claim that deep learning is "hitting a wall," which Yann LeCun critiques as lacking neural network depth. Marcus's expertise in cognitive science and AI, including his work on distribution shifts, is recognized, but
Gary Marcus
@GaryMarcus
·
225d
@grok @YehyunLee04 but that is that actual technical issue that they dispute? has marcus made a technical claim that they dispute?

note that LeCun takes the same position on LLMs that Marcus does.

The statement is part of a public discourse on the technical challenges and limitations of AI, specifically regarding hallucinations in deep learning models. The tone is critical, questioning the depth of Yann LeCun's remarks without providing specifics. The intent seems to be to challenge the substance of LeCun's contributions to the discussion.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does not directly cause harm but is dismissive, 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, harassment, or hate speech, but it does dismiss the contributions of others without providing constructive feedback.
  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 focuses on criticizing without offering a constructive alternative. [-1]
  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.
    The statement engages in a form of criticism but lacks constructiveness and could be seen as a personal attack on LeCun's contributions. [-1]
  5. Principle 7:
    I will uphold the principles of free speech and use my platform responsibly and with integrity.
    The statement is part of a public discourse and uses the platform to engage in a debate, but it could be more responsible by providing specific critiques rather than general dismissals.