Gary Marcus

Rank 14 of 47
|
Score 92
4/8/2024, 10:52:21 PM
X
In reply to:
Anonymous746292
@Anonym64382296
·
536d
@GaryMarcus @Anders_Research @geoffreyhinton But this is superhuman ability, especially since it can do this for a massive number of subjects. How many humans get a correct answer 90% of the time even after studying (the human training set)?
Gary Marcus
@GaryMarcus
·
536d
@Anders_Research @geoffreyhinton For cases in the training set, sure
Anders Research
@Anders_Research
·
536d
@GaryMarcus @geoffreyhinton You are losing me Gary… Chat-GPT gives the right answers 90%+ of the time. The spirit of this Hinton quote is obviously true.
Gary Marcus
@GaryMarcus
·
536d
Egregiously misleading rhetoric from @geoffreyhinton.

LLMs can “answer more less any question” but “in a sensible way” is utterly misleading; it abundantly clear that LLMs regularly hallucinate and make boneheaded errors, in essentially every domain that’s been explored.

(Also…

The statement challenges the notion that current AI systems have achieved a level of general intelligence comparable to humans by pointing out that benchmarks may not reflect real-world tasks. It calls for a demonstration of AI performing complex tasks such as driving a car or playing a new video game, which are activities that require a broad range of cognitive abilities.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does not appear to cause harm with its words or actions. [+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 and does not engage in cyberbullying, harassment, or hate speech. [+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 promotes understanding by highlighting the difference between benchmark performance and real-world tasks. [+1]
  4. Principle 5:
    I will acknowledge and correct my mistakes.
    The statement engages in constructive criticism by challenging the benchmarks used to evaluate AI systems. [+1]
  5. Principle 6:
    I will use my influence for the betterment of society.
    The statement corrects a perceived overestimation of AI capabilities, contributing to a more accurate understanding of the current state of AI. [+1]
  6. Principle 7:
    I will uphold the principles of free speech and use my platform responsibly and with integrity.
    The statement uses the platform responsibly to engage in a substantive discussion about AI capabilities. [+1]