Gary Marcus

Rank 12 of 47
|
Score 83
2/14/2025, 12:25:01 AM
X
In reply to:
Gary Marcus
@GaryMarcus
·
342d
Crossroads in AI history:

Combined with yesterday’s implicit concession by Altman that GPT-5 cannot be built by pure scaling alone, @melmitchell1’s new detailed look (links below) at LLMs and claims of “emergent” models of the world is deeply damaging to the ambitions of…
Melanie Mitchell
@MelMitchell1
·
342d
Do large language models develop "emergent" models of the world? My latest Substack posts explore this controversial claim & more generally the nature of "world models":

LLMs & World Models, Part 1: https://aiguide.substack.com/p/l…

LLMs & World Models, Part 2: https://aiguide.substack.com/p/l…

The statement 'Half a *trillion*, that is' is a brief, somewhat sarcastic remark likely referring to the scale or cost associated with developing large language models (LLMs) like GPT-5. It is a response to a discussion about the limitations and ambitions of AI models, particularly in the context of their development and potential capabilities. The tone suggests skepticism or criticism of the resources required for such AI advancements.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does not directly cause harm but implies criticism of the resources involved in AI development.
  2. 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 does not engage in constructive criticism or dialogue; it is more of a sarcastic remark without substantive engagement. [-1]
  3. Principle 7:
    I will uphold the principles of free speech and use my platform responsibly and with integrity.
    The statement uses free speech to express skepticism but lacks depth or responsibility in addressing the issue constructively. [-1]