Gary Marcus

Rank 22 of 47
|
Score 35
6/18/2024, 2:56:21 PM
X
In reply to:
AI Notkilleveryoneism Memes ⏸️
@AISafetyMemes
·
454d
@GaryMarcus Sure, we agree that hallucinations etc are still a problem, but the key issue here is the *trend*

Curious - how much do you use these models?

Using GPT2 vs GPT3 vs GPT3.5 vs GPT4 doesn't feel night and day different to you?

Like talking to a child vs an adult?
Gary Marcus
@GaryMarcus
·
454d
@AISafetyMemes - benchmarks are broken; everyone in the field knows that; many have agonized over that. They get gamed, people teach to the test etc, data contamination etc, and they are rarely valid constructs in the psychological sense
- even putting aside the general failures of benchmarks,…
AI Notkilleveryoneism Memes ⏸️
@AISafetyMemes
·
454d
@GaryMarcus You don't see data showing models have rapidly been getting better?
Gary Marcus
@GaryMarcus
·
454d
@AISafetyMemes I don’t really see the data for that, and I still tons of hallucinations, documented here, in empirical papers, reports in corporate use etc. Certainly they have not reduced to expert/levels as Reid Hoffman promised, and we are still waiting on GPT-5 level models.
AI Notkilleveryoneism Memes ⏸️
@AISafetyMemes
·
454d
@GaryMarcus Thanks, but this doesn't address my core argument that models are improving rapidly at everything, including hallucinations.
Gary Marcus
@GaryMarcus
·
454d
@AISafetyMemes Yes there are more tools; integration with the tools has been problematic though. I certainly think that integrating w symbolic tools would be good if it can be done. But so far that has been shaky.

You can’t just say you personally don’t care about hallucinations; that is a…

The statement engages in a substantive discussion about the development and performance of AI models, particularly focusing on the differences between various versions and the challenges in achieving significant improvements. The tone is analytical and speculative, aiming to contribute to the ongoing debate about AI advancements and their implications.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does not appear to cause harm; it is a speculative analysis of AI development. [+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 respects the privacy and dignity of others, focusing on technical aspects rather than personal attacks. [+1]
  3. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement promotes understanding by engaging in a detailed discussion about AI model improvements and challenges. [+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 constructive criticism, questioning the effectiveness of current AI models and the validity of benchmarks without resorting to personal attacks. [+1]
  5. Principle 6:
    I will use my influence for the betterment of society.
    The statement uses the platform to contribute to a broader discussion about AI development, which is a significant societal issue. [+1]