Gary Marcus

Rank 19 of 47
|
Score 64
6/29/2025, 5:00:38 PM
X
In reply to:
Chris
@chatgpt21
·
10d
@JimDean9000 @GaryMarcus Because llms don’t work the same way as human brains do? Yet, they can still do superhuman things with 100% accuracy.
Jim Dean
@JimDean9000
·
10d
@GaryMarcus @chatgpt21 Can we take a step back and see how fucked up this is? My 9 yo niece can play chess following the rules, of course she's rubbish and I'd win if I played seriously with her, but I had to tell her the rules twice. Think about it: how can this tech lead to AGI? It doesn't compute.
Gary Marcus
@GaryMarcus
·
10d
@chatgpt21 Chris, I honestly and truly cannot believe how dumb this tweet is, on multiple levels. One is that you OBVIOUSLY didn’t read the paper that you are criticizing.

The other is this: If GPT-5 can’t play chess without making illegal moves 5% of the time – despite exposure to
Chris
@chatgpt21
·
10d
Gary Marcus can’t go 5 minutes without saying o3 can’t play chess.

Eventually it will be “GPT 5 can play chess but it made an illegal move 5% of the time”

The statement is part of a broader discussion on the capabilities of large language models (LLMs) and their comparison to human cognition. It addresses the limitations and potential of LLMs in achieving artificial general intelligence (AGI). The tone is neutral and factual, comparing LLMs to calculators in terms of their capabilities.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does no harm and is a factual comparison, adhering to the principle. [+1]
  2. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    It promotes understanding by providing a comparison that can help clarify the capabilities of LLMs. [+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 dialogue without personal attacks, adhering to the principle. [+1]