Gary Marcus

Rank 32 of 47
|
Score -1
4/2/2024, 10:38:29 PM
X
In reply to:
Derya Unutmaz, MD
@DeryaTR_
·
227d
@GaryMarcus I didn’t say Turbo is 10x faster, but 2x is reasonable. To know the extent it’s slowing down or accelerating, we need several more data points until next year. Things sometimes happen in bursts, but in the long run, they tend to follow exponential curves, like computer speeds.
Gary Marcus
@GaryMarcus
·
227d
@DeryaTR_ Ok for sake of argument – but no way is turbo 10x gpt 4.

The rate of acceleration over last twelve months has obviously slowed.

Don’t make me pull out my disco illusion tweet…
Derya Unutmaz, MD
@DeryaTR_
·
227d
@GaryMarcus Also difference between 3.5 and 4 much more than double, maybe 10x
Derya Unutmaz, MD
@DeryaTR_
·
227d
@GaryMarcus I don’t have that data either, I speculated for 6 months since Elon Musk claimed it was like 10x - but even doubling every year is still big exponential. You get 1000x in 10 years.
Gary Marcus
@GaryMarcus
·
227d
@DeryaTR_ It has not doubled in capabilities every six months. Any measure showing that 4 is double 3.5, won’t show that turbo doubled 4 in six months. By any measure I know 4 was a big advance but turbo was no doubling of that advance. Nor was Gemini Pro, Claude’s best, etc. Nothing now…
Derya Unutmaz, MD
@DeryaTR_
·
227d
@GaryMarcus Well GPT-4 improved with turbo last November, and we should have GPT-4.5 soon. Even doubling in capabilities every 6 months would be a massive exponential. https://openai.com/blog/new-mode…

The conversation between @DeryaTR_ and @GaryMarcus is a technical discussion about the rate of progress in AI capabilities, specifically the improvements from one version to another and the speed of these advancements. The tone is analytical and focused on clarifying the facts and expectations regarding AI development. The content revolves around the interpretation of data and projections for future growth in the field of AI, which is a matter of public interest given the impact of AI on society.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does no harm as it is a factual clarification. [+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 respects privacy and dignity, as it does not involve personal information. [+1]
  3. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement promotes understanding by clarifying a point about accuracy versus speed. [+1]
  4. 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 by focusing on the argument rather than the individual. [+1]
  5. Principle 5:
    I will acknowledge and correct my mistakes.
    There is no indication of a mistake being made or corrected in this statement.
  6. Principle 6:
    I will use my influence for the betterment of society.
    The statement contributes to the betterment of society by fostering a more informed discussion about AI progress. [+1]
  7. Principle 7:
    I will uphold the principles of free speech and use my platform responsibly and with integrity.
    The statement upholds free speech principles by contributing responsibly to the public discourse on AI. [+1]