Demis Hassabis: Transfer learning is key to AGI
00:00:00.000 | So transfer learning is when you transfer your knowledge from one domain to a new,
00:00:04.720 | totally new domain.
00:00:05.800 | And I think that is the key to actually general intelligence.
00:00:08.560 | And that's the thing that we as humans do amazingly well.
00:00:11.240 | For example, I played so many board games now.
00:00:14.440 | Uh, when I, if someone was to teach me a new board game, I wouldn't be coming to that
00:00:17.840 | fresh anymore.
00:00:18.440 | I would know, you know, straight away I could apply all of these different heuristics to
00:00:22.520 | that I've learned from other games to this new one, even if I've never seen that before.
00:00:25.920 | And currently no machines can do that.
00:00:28.600 | And I think the key to doing transfer learning is going to be a conceptual knowledge.
00:00:33.080 | So abstract knowledge.
00:00:34.760 | So the, so the acquisition of conceptual knowledge that is abstracted away from the
00:00:40.280 | perceptual details of where you learned it from.
00:00:42.760 | So then you can go, okay, I'll apply it to this new domain.
00:00:45.760 | Um, so I think that's actually one of the, the, the big challenges to be tackled towards
00:00:51.080 | a general AI.