Gary Marcus

Rank 11 of 47
|
Score 114
6/29/2025, 12:12:37 AM
X
In reply to:
Juan I. Fernandez
@juanif17
·
46d
@stanmaltman @GaryMarcus GPT is already better than most professors, editors and many other employees

Even without understanding shit
Stan Maltman (parody)
@stanmaltman
·
47d
@juanif17 @GaryMarcus Well evidently it is quite difficult.
Juan I. Fernandez
@juanif17
·
47d
@GaryMarcus We dont need LLMs to understand it, we need them to behave as if they did

Plus, the average human dont understand it...we just need the bar the be slightly higher to be better than the average joe

not that difficult
Gary Marcus
@GaryMarcus
·
47d
BREAKING: Explosive new paper from MIT/Harvard/UChicago.

Things just got worse — a lot worse — for LLM’s and the myth that they can understand and reason.

The paper documents a pattern they called Potemkins, a kind of reasoning inconsistency (see figure below). They show that

The statement is part of a discussion on the capabilities of large language models (LLMs) and their understanding, sparked by a paper highlighting reasoning inconsistencies in LLMs. The reply 'says more about you than them, IMHO' is a personal remark directed at a participant in the conversation, suggesting that their opinion reveals more about their own biases or perspectives than about the subject being discussed.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does not directly harm but implies a personal judgment, which could be seen as dismissive. [-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 does not respect the dignity of others, as it shifts focus from the topic to a personal critique. [-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 does not engage in constructive dialogue and instead makes a personal remark, which could hinder understanding. [-1]