korrents

korrents · video

Dario Amodei (Anthropic CEO) — The hidden pattern behind every AI breakthrough

Dario Amodei · 1h 58m · youtube.com ↗

30 korrents from this recording

1h
Dario Amodei did not write this page.

Every claim below is a statement made in this recording, quoted word for word and linked to the second it was said, so you can hear it rather than take our word for it. The wording comes from the transcript published alongside the recording; the sentence above each quote is our reading of the claim, not their wording.

  1. 0:03:35 · watch on youtube.com↗

    It's predictable to sometimes even several significant figures, which you don't see outside of physics. You don't expect to see it in this messy empirical field. But actually, specific abilities are very hard to predict.
  2. 2 min later
  3. 0:05:18 · watch on youtube.com↗

    I definitely think that things like alignment and values are not guaranteed to emerge with scale. One way to think about it is you train the model and it is basically predicting the world.
  4. 2 min later
  5. 0:07:22 · watch on youtube.com↗

    I think we're already at the point where The things the models can't do don't seem to me to be different in kind from the things they can do.
  6. 3 min later
  7. 0:10:11 · watch on youtube.com↗

    There's many sources of data in the world and there's many ways that you can also generate data. My guess is that this will not be a blocker.
  8. 4 min later
  9. 0:13:46 · watch on youtube.com↗

    The models just want to learn. You get the obstacles out of their way. You give them good data. You give them enough space to operate in. You don't do something stupid like condition them badly numerically and they want to learn.
  10. 6 min later
  11. 0:19:31 · watch on youtube.com↗

    Stepping back from it, one of the reasons why I'm very empiricist about AI, about safety, about organizations is that you often get surprised.
  12. 2 min later
  13. 0:21:11 · watch on youtube.com↗

    And so I don't know, it turns out that intelligence isn't a spectrum. There are a bunch of different areas of domain expertise. There are a bunch of different kinds of skills.
  14. 3 min later
  15. 0:23:46 · watch on youtube.com↗

    I think my view is we're going to keep getting better across the board. And I don't see any area where the models are super weak or not starting to make progress.
  16. 4 min later
  17. 0:27:58 · watch on youtube.com↗

    In terms of someone looks at the model and even if you talk to it for an hour or so, it's basically like a generally well-educated human. Yeah. That could be not very far away at all, I think. That could happen in 2 or 3 years.
  18. 2 min later
  19. 0:30:09 · watch on youtube.com↗

    Another thing that could be the case is that I think there are these mysterious frictions that don't show up in naive economic models, but you see it whenever you go to a customer or something
  20. 2 min later
  21. 0:31:57 · watch on youtube.com↗

    But I'm skeptical of any precise mathematical or exponential prediction of how it's going to be. I think it's all going to be a mess, but I think what we know is it's on a metaphorical exponential and it's going to happen fast.
  22. 9 min later
  23. 0:40:27 · watch on youtube.com↗

    But we saw enough signs of the models doing those key things well. And if we look at state-of-the-art models and go backwards to previous models, we look at the trend, it shows every sign of 2 or 3 years from now, We're going to have a real problem.
  24. 7 min later
  25. 0:47:03 · watch on youtube.com↗

    I think compartmentalization is the best way to do it. Just limit the number of people who know about something. If you're a 1,000-person company and everyone knows every secret, one, I guarantee you have a leaker, and two, I guarantee you have a spy, like a literal spy.
  26. 1 min later
  27. 0:47:59 · watch on youtube.com↗

    all the current methods that involve some kind of fine-tuning, of course, have the property that the underlying knowledge and abilities that we might be worried about don't disappear. The model is just taught not to output them.
  28. 9 min later
  29. 0:57:11 · watch on youtube.com↗

    And we think that we have a very high talent density of folks who are good at that. And my view has always been talent density beats talent mass.
  30. 3 min later
  31. 0:59:50 · watch on youtube.com↗

    I just feel like you discover 10 new ideas and 10 new ways that things can go wrong by trying these in practice. And that empirical learning, I think it's just not as widely understood as it should be.
  32. 1 min later
  33. 1:01:02 · watch on youtube.com↗

    you see this phenomenon over and over again where it's like the scaling and the safety are these 2 snakes that are coiled with each other, always even more than you think.
  34. 4 min later
  35. 1:04:54 · watch on youtube.com↗

    Any plan that actually succeeds, regardless of how hard misalignment is to solve, any plan that actually succeeds, is going to need to solve misuse as well as misalignment.
  36. 2 min later
  37. 1:06:27 · watch on youtube.com↗

    There needs to be some kind of legitimate process for managing this technology, which includes the role of the people building it, includes the role of democratically elected authorities, includes the role of all the individuals who will be affected by it.
  38. 5 min later
  39. 1:10:57 · watch on youtube.com↗

    I think in general, unitary visions for what it means to live a good life have not worked out well at all.
  40. 2 min later
  41. 1:12:51 · watch on youtube.com↗

    My worry is that if the basic incentives are about national security and power, that's going to become clear sooner or later. If they see this as a source of national power, they're going to at least try to do what's most effective. And that could lead them in the direction of AGI.
  42. 3 min later
  43. 1:15:25 · watch on youtube.com↗

    I think there's a general feeling that models are misaligned or there's an alignment problem to solve, kind of like the Riemann hypothesis or something. Someday we'll crack the Riemann hypothesis. I don't quite think it's like that.
  44. 2 min later
  45. 1:17:07 · watch on youtube.com↗

    And so I actually think fact number 1 and fact number 2 are enough to be really worried. You don't need all this detailed stuff about convergent instrumental goals or analogies to evolution.
  46. 2 min later
  47. 1:19:02 · watch on youtube.com↗

    I don't think we're aligned by default. I don't think we're doomed by default and have some problem we need to solve. It has some kind of different character.
  48. 4 min later
  49. 1:22:58 · watch on youtube.com↗

    I have a substantial probability mass on this all goes wrong. It's a complete disaster, but in a completely different way than anyone had anticipated.
  50. 5 min later
  51. 1:27:58 · watch on youtube.com↗

    I don't think there should be one constitution for a model that everyone uses. Probably models' constitutions should be very simple. It should only have very basic facts that everyone would agree on.
  52. 6 min later
  53. 1:33:38 · watch on youtube.com↗

    I think a good norm would be people see these cybersecurity leaks from companies or leaks of the model parameters or something and say, they screwed up. That's bad. If I'm a safety person, I might not want to work there.
  54. 14 min later
  55. 1:47:21 · watch on youtube.com↗

    So it's kind of a race between how fast the technology is getting better and how fast it's integrated into the economy. I think that's just a very unstable and turbulent process.
  56. 4 min later
  57. 1:51:40 · watch on youtube.com↗

    the more we've looked at these language models and particularly looked inside them to see things like induction heads, a lot of the cognitive machinery that you would need for active agents seems already present in the base language models.
  58. 5 min later
  59. 1:56:46 · watch on youtube.com↗

    just attaching your incentives very strongly to the approval or cheering of a crowd. I think that can destroy your mind and in some cases it can destroy your soul.