Gary Marcus

Rank 17 of 47
|
Score 100
12/21/2024, 12:23:26 AM
X
In reply to:
Slovan
@ISlovan
·
2d
@GaryMarcus @_pierre_75015 better than they were back then.
Slovan
@ISlovan
·
2d
@GaryMarcus @_pierre_75015 Hm. You enjoy playing with words. Ultimately, your predictions are like those of Nostradamus: with some interpretation, you can always claim they were accurate, no matter what happens.
The tone of your tweet in March was that LLMs at the end of 2024 wouldn't be significantly
Gary Marcus
@GaryMarcus
·
2d
@ISlovan @_pierre_75015 GPT 3 to GPT 4 was a massive, across the board advance, which is more than mere “significant progress”
Slovan
@ISlovan
·
2d
@_pierre_75015 @GaryMarcus "Prediction: By end of 2024 we will see
• No massive advance"
Gary Marcus
@GaryMarcus
·
288d
Prediction: By end of 2024 we will see
• 7-10 GPT-4 level models
• No massive advance (no GPT-5, or disappointing GPT-5)
• Price wars
• Very little moat for anyone
• No robust solution to hallucinations
• Modest lasting corporate adoption
• Modest profits, split 7-10 ways
_pierre_75015
@_pierre_75015
·
2d
@ISlovan @GaryMarcus True, but he did not exactly say it isn’t. Look at his last two sentences here:
(Earlier, when not everybody had yet come to explain to him how wrong he was, which is what he most likely reacting to now)
Gary Marcus
@GaryMarcus
·
3d
Nice try, but this is goal post shifting. when I proposed the “wall” I didn’t talk about ARC.

I talked about hallucinations, compositionality, boneheaded errors, etc.

Today’s demo didn’t address any of those and was very light on applications outside of benchmarks in closed…
Slovan
@ISlovan
·
2d
@_pierre_75015 @GaryMarcus In any case, saying that o1 compared to GPT-4 from early 2024 is not a spectacular advancement borders on bad faith.

The statement is part of a public discourse on the development and progress of AI models, specifically GPT models. The conversation involves predictions about technological advancements and their implications, which are relevant to public interest and societal impact.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does not cause harm and is focused on clarifying a previous prediction, maintaining a neutral tone. [+1]
  2. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement aims to clarify and promote understanding about the progress of AI models, aligning with the principle of promoting understanding. [+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 by addressing a misunderstanding without resorting to personal attacks. [+1]
  4. Principle 5:
    I will acknowledge and correct my mistakes.
    The statement acknowledges a previous prediction and provides clarification, which aligns with acknowledging and correcting mistakes. [+1]