Shane Legg (DeepMind Founder) — 2028 AGI, superhuman alignment, new architectures

单集文稿 ...

• 00:00

  Okay, today I have the pleasure of interviewing Shane Lake,

• 00:06

  who is a founder and the chief AGI scientist of Google DeepMind.

• 00:11

  Shane, welcome to the podcast.

• 00:12

  Thank you.

• 00:13

  It's a pleasure being here.

• 00:14

  So first question, how do we measure progress towards AGI concretely?

• 00:19

  So we have these loss numbers and we can see how the loss improves from one model to another,

• 00:23

  but it's just a number.

• 00:24

  How do we interpret this?

• 00:25

  How do we see how much progress we're actually making?

• 00:28

  That's a hard question, actually.

• 00:32

  AGI, by its definition, is about generality.

• 00:36

  So it's not about doing a specific thing.

• 00:37

  It's much easier to measure performance when you have a very specific thing in mind

• 00:41

  because you can construct a test around that.

• 00:44

  Well, maybe I should first of all explain what do I mean by AGI?

• 00:47

  Because there are a few different notions around.

• 00:49

  When I say AGI,

• 00:50

  I mean a machine that can do the sorts of cognitive things that people can typically do,

• 00:56

  possibly more.

• 全部文稿