Gary Marcus

Rank 37 of 47
|
Score -82
6/26/2024, 1:37:08 AM
X
In reply to:
Igor Kurganov
@IgorKurganov
·
129d
@GaryMarcus I’d take the bet you propose for AI.

Or, in the spirit of the OP, I’d bet on multimodal LLMs hitting the majority of 1, 2, 5 within four years.

Since we’re comparing to human capabilities, I’m sure you agree that e.g. the arbitrariness of the kitchen ought to be constrained.
Gary Marcus
@GaryMarcus
·
130d
@IgorKurganov Discussed here
Gary Marcus
@GaryMarcus
·
131d
People keep complaining that there is no definition and criteria for AGI.

That’s not true.

I actually gave some in 2023, in a bet I proposed to Elon Musk, & vetted with the community, including two of people that helped coined the term, and *there was a large degree of…
Igor Kurganov
@IgorKurganov
·
130d
@GaryMarcus My claim is that I think skeptics, including you, are likely wrong about specific abilities they expect LLMs to not have in 2 or 4 years.

Given that I say an LLM probably can do it, it makes sense for you to point at an example you think it won’t be able to do in 2 or 4 years.
Gary Marcus
@GaryMarcus
·
130d
@IgorKurganov I am not seeing the bet or the claim as of yet, and would note that some studies show quality and security of code is decreasing.
Igor Kurganov
@IgorKurganov
·
130d
@GaryMarcus Coding has improved a lot over the last two years and that’s without incoming much larger training runs.

Happy to structure a bet if we can find specific LLM abilities we disagree on whether they exist in 2 or 4 years.

Which challenges humans do on a computer come to mind?
Gary Marcus
@GaryMarcus
·
130d
@IgorKurganov It might have one or two more, but even one more doubling is somewhat in doubt. What’s your measure? Seems like nothing is really noticeably better than GPT-4 in almost two years, despite huge investments, and costs (data, $, environmentally) themselves may be increasing…

The statement and the preceding conversation constitute public discourse as they engage in a substantive debate about the future capabilities of AI, specifically large language models (LLMs), and their potential to perform tasks comparable to human abilities. The discussion involves predictions, challenges, and the setting of benchmarks for AI development, which are relevant public issues in the field of technology and ethics.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does not cause harm and is part of a constructive debate about AI capabilities. It seeks clarification and further discussion on the constraints of the proposed bet, which is a healthy part of public discourse. [+1]
  2. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement promotes understanding by seeking to clarify the terms of the bet and the constraints on the 'arbitrary kitchen' scenario. This helps ensure that the discussion is based on clear and agreed-upon terms, fostering a more informed and empathetic dialogue. [+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 engages in constructive criticism and dialogue by questioning the omission of certain points and seeking clarification. It does not engage in personal attacks or ad hominem arguments, maintaining a respectful tone. [+1]