12
Pacing the Frontier: A statement from 1,132 employees of frontier AI companies (www.pacingthefrontier.com)
submitted 3 months ago by to c/technology@lemmy.world
24 comments fedilink hide all child comments

all 24 comments

sorted by: hot top controversial new old
[–] 2 points 3 months ago (1 child)

I felt like frontier labs need to teach everybody how to do alignment research without math. Speed up research because those who in power never slow down their "strategic advantages".

  • permalink
  • fedilink
  • source
  • hideshow 1 child comment
    •
  • [–] [S] 1 point 3 months ago

    That’s a more realistic ask than hoping everyone slows down: If alignment methods can be made accessible to engineers without deep math, more people can stress test, critique and improve them. Knowledge spreads far more easily than voluntary restraint, especially when geopolitical incentives point the other way.

  • permalink
  • fedilink
  • source
  • parent
    • [–] 2 points 3 months ago

    Stopping the development of larger and larger models because of the unmanageable environmental cost it brings: ❌
    Slowing down AI development because of fear of AGI: ✔️

  • permalink
  • fedilink
  • source
    •
    [+] [S] -8 points 3 months ago (20 children)

    »Slow down AI« sounds noble until you ask who actually slows down: Democracies and companies that comply simply hand the advantage to rivals that don’t. History suggests voluntary technological restraint rarely wins against geopolitical competition. The race doesn’t stop: it just changes leaders.

  • permalink
  • fedilink
  • source
  • hideshow 20 child comments
    •
  • [–] 5 points 3 months ago (6 children)

    This all presupposes that this “arms race” is going somewhere. Far as I can tell, trillions of dollars will be incinerated making a better word-guesser while claiming it will lead to some nebulous “AGI”.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 6 child comments
    •
  • [–] 1 point 3 months ago

    You’re both right and wrong. There is an arms race with AI and it is going somewhere. The problem is that its destination isn’t so obvious to Americans, because the US is already loosing. The US is looking toward a model for AI that isn’t sustainable, cheap subscriptions to the highest possible tier models. China is looking toward a model of deep integration. Chinese AI is helping robots balance, drone swarms coordinate, it’s helping factories go dark… They’re even open sourcing some of the best models available (e.g., Kimi 3). In the US, you have companies hoping they can somehow ban competing Chinese models.

  • permalink
  • fedilink
  • source
  • parent
  • [–] [S] 0 points 3 months ago (4 children)

    How much more effective have the large models become for you over the last three years? Do you think development might come to a halt in China, for example, if US providers were indeed to stop developing or releasing them?

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 4 child comments
    •
  • [–] 1 point 3 months ago (3 children)

    I think they’ve hit the point of diminishing returns and want to frame it as “we’d better stop before we all die”. That is, incremental improvement is all they’re getting from throwing exponentially increasing compute at these things.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 3 child comments
    •
  • [–] [S] 0 points 3 months ago (2 children)

    So have they been getting better and better?

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •
  • [–] 6 points 3 months ago* (last edited 3 months ago) (12 children)

    Ah, yes. We can't stop because the 'bad guys' won't. God forbid anyone win the technological arms race other than the land of the once free and the now cowed. If another country wins global dominance they might treat the US the way the US has treated the rest of the world, and that would be unthinkable.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 12 child comments
    •
  • [–] 3 points 3 months ago* (10 children)

    I don't think @eicker put all that much thought into that comment lol

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 10 child comments
    •
  • [–] 3 points 3 months ago (8 children)

    Please don't waste your time using AI powered 'AI detectors'. There's no such thing. Some relatively rare AI generated images and videos contain a synthid or other tag to show that they were AI generated, but those can always be removed and it's not something you can do with plain text.

    There is absolutely zero benefit to using something like this, but it can cause harm to real people as these often falsely report that an author or student's actual own words are AI generated.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 8 child comments
    •
  • [–] 0 points 3 months ago* (4 children)

    Yea it's a statistical guess. The statistics are weighted in a certain direction for me seeing how many times this person seems to be caught up in the same accusation and all of his profile link websites looking... the way it does. But of course it could just be that he happens to write a lot like how AI would generate text.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 4 child comments
    •
  • [–] 2 points 3 months ago* (2 children)

    It's a blind guess that's weighted against people who are well read and don't type like complete idiots. AI are trained on human language. They write like human writers. There's no universal magic formula for "how AI would generate text." AI detectors are nonsense. That is all they are. The only thing you're doing by using them is feeding data in and accusing people of being witches if an AI tells you they are.

    Aha! Now I've got you!

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •
  • [–] [S] -2 points 3 months ago

    It’s not about who deserves to win, it’s about incentives. If one side unilaterally slows development while others don’t, the likely outcome is a shift in technological leadership, not a slower pace overall. Whether that’s the US, China, or anyone else is a separate political question from whether voluntary restraint is an effective strategy.

  • permalink
  • fedilink
  • source
  • parent