Gary Marcus

Rank 11 of 47
|
Score 114
9/13/2024, 3:13:06 PM
X
In reply to:
Leon Lang
@Lang__Leon
·
333d
@GaryMarcus @tegmark My question to you @GaryMarcus : do you agree with that prediction? If not, why not? If yes, then what is the crux?
Leon Lang
@Lang__Leon
·
333d
@GaryMarcus @tegmark … and which are as simple as playing tic-tac-toe. My very strong prediction is that if you’d evaluate successively more capabilities on your private benchmark of such tasks, you would, in fact, see improvements.
Leon Lang
@Lang__Leon
·
333d
@GaryMarcus @tegmark To be extra charitable, I could interpret you as saying “if I’d create the benchmark, people would just game that benchmark, which proves nothing.”
If that’s the case, then I think I’d advice you to create a private benchmark with just 20 questions current models do poorly on…
Leon Lang
@Lang__Leon
·
333d
@GaryMarcus @tegmark Unless, of course, you think the benchmarks are missing something important. But that should make it possible to create a benchmark where capabilities do not improve. I think no one managed to create such a benchmark.
Leon Lang
@Lang__Leon
·
333d
@GaryMarcus @tegmark I don’t understand the appeal of pointing to a lack of specific capabilities like “play tic-tac-toe”. As long as there is progress on benchmarks, capabilities increase, and should eventually also include tic-tac-toe.
Gary Marcus
@GaryMarcus
·
333d
@tegmark Max @tegmark you are just giving more power to the companies you rightly want to regulate by playing into their hype.

The new system can’t even play chess or tic-tac-toe reliably.

The statement 'chess was one of my private benchmarks. oops.' is part of a larger conversation about the capabilities of AI systems and their benchmarks. This conversation involves public figures discussing the progress and limitations of AI, which is a substantive public issue. Therefore, it constitutes public discourse.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does not appear to cause harm with words or actions. It is a lighthearted admission of a private benchmark, which does not seem harmful. [+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. It does not engage in cyberbullying, harassment, or hate speech. [+1]
  3. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement does not explicitly promote understanding, empathy, or compassion, but it does contribute to the dialogue in a non-hostile manner.
  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 is part of a constructive dialogue and does not engage in personal attacks or ad hominem arguments. It acknowledges a mistake in a lighthearted way. [+1]
  5. Principle 5:
    I will acknowledge and correct my mistakes.
    The statement acknowledges a mistake ('oops'), which aligns with the principle of acknowledging and correcting mistakes. [+1]
  6. Principle 6:
    I will use my influence for the betterment of society.
    The statement uses the influence of the speaker to contribute to a public discussion about AI capabilities, which can be seen as using influence for the betterment of society. [+1]
  7. Principle 7:
    I will uphold the principles of free speech and use my platform responsibly and with integrity.
    The statement upholds the principles of free speech and uses the platform responsibly and with integrity. [+1]