Gary Marcus

Rank 11 of 47
|
Score 114
6/15/2025, 3:41:08 PM
X
In reply to:
Ryan Greenblatt
@RyanPGreenblatt
·
59d
@GaryMarcus Yes, concretely Daniel K updated his medians to be 1 year later (an approximately 40% increase at the time!). I can dig up a source if you want.

I've also update towards somewhat longer timelines based on reduced expectations for pretraining.
Gary Marcus
@GaryMarcus
·
59d
@RyanPGreenblatt ok fair, i overstated a bit.

but look the whole field though pure pretraining would do well and then in November many reports emerged saying that recent models like Orion (relabeled as 4.5) and Behemoth didn’t meet expectations. And GPT-5 still hasn’t emerged.

Did that change
Ryan Greenblatt
@RyanPGreenblatt
·
59d
@GaryMarcus No, see the prior 2 bullets for other things that would cause me to update.

I was just listing some of the things that would cause me to update (as you claimed that nothing would cause update), of course in practice there are many types of observations that would cause updates.
Gary Marcus
@GaryMarcus
·
59d
@RyanPGreenblatt but wait, you aren’t really going to slow your timelines that say AGI is fairly like to arrive in next few years until 2032 arrives?
Ryan Greenblatt
@RyanPGreenblatt
·
59d
@GaryMarcus Sure, listing things that would cause me to extend my timelines:

- Progress on completing longer (in human equivalent duration) software engineering tasks slows down either in benchmarks or possibly in real world usefulness. Or generally progress on software engineering/
Gary Marcus
@GaryMarcus
·
59d
1. What would convince me to shorten my AGI timelines? robust, principled solutions to things like hallucinations and stupid errors.

2. What would convince doomers to extend their timelines? As far as I can tell, nothing. (Prove me wrong?)

The statement is part of a public discourse on the topic of artificial general intelligence (AGI) timelines, involving a discussion on updates and qualifications in the field. The tone is respectful and acknowledges the credit due to individuals for updating their views based on new information.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does no harm and acknowledges the positive actions of others, aligning with the principle of doing no harm. [+1]
  2. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement promotes understanding by recognizing the importance of updates and qualifications in public discourse, aligning with the principle of promoting understanding and empathy. [+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 acknowledging the updates and qualifications made by others, without resorting to personal attacks, aligning with the principle of engaging in constructive criticism. [+1]