Gary Marcus

Rank 18 of 47
|
Score 63
4/21/2024, 6:40:13 PM
X
In reply to:
Dagan Shani
@daganshani1
·
422d
@Widdrat @GaryMarcus Note that Claude is the one who initiated admission to fabricate the answer (which it did), not the user. This indicates that Claude knows it cheated. How does this qualify as a hallucination?
Nostalgisch im Lidl
@Widdrat
·
422d
@daganshani1 @GaryMarcus The "admitting of fabrication" is a hallucination in and of itself.
Dagan Shani
@daganshani1
·
422d
@GaryMarcus Actually Claude 2 admitted it fabricated stuff before:
Dagan Shani
@daganshani1
·
702d
Claude 2 admits it fabricated a person(A CEO of a Biotech Company), apologizes and shares its conclusions and self reflection about the cheating with the user(a reporter friend). Please see the original admission and apology(in Mandarin)+English translation in the pictures:
Gary Marcus
@GaryMarcus
·
422d
So many people are confused about the relation between human cognitive errors and LLM hallucinations that I wrote this short explainer:

Humans say things that aren't true for many different reasons
• Sometimes they lie
• Sometimes they misremember things
• Sometimes they fail…

The statement in question is part of a broader discussion about the capabilities and limitations of AI, specifically addressing the issue of AI-generated content and its potential to mislead or 'hallucinate'. The conversation revolves around the nature of AI responses, whether they can be considered as 'knowing' or intentional fabrications, and the implications of such behavior.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement contributes to understanding AI behavior, which is crucial for users and developers. It clarifies that AI does not 'know' or intentionally lie but responds based on programmed algorithms and available data. [+2]
  2. Principle 2:
    I will respect the privacy and dignity of others and will not engage in cyberbullying, harassment, or hate speech.
    The statement respects the dignity of the discourse by focusing on the technical aspects of AI behavior without devolving into personal attacks or derogatory language. [+1]
  3. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    By explaining AI limitations, the statement fosters a more informed and empathetic understanding among the audience about technology's current state and its potential pitfalls. [+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.
    The statement engages constructively by responding to a previous claim, adding clarity and depth to the discussion, which encourages more informed dialogue. [+2]
  5. Principle 6:
    I will use my influence for the betterment of society.
    The statement uses the platform to educate and inform, which ultimately serves the betterment of society by fostering a more accurate understanding of AI technologies. [+2]