training ai without writing a reward function, with reward modelling
Published 4 years ago • 239K plays • Length 17:52Download video MP4
Download video MP3
Similar videos
-
10:22
ai that doesn't try too hard - maximizers and satisficers
-
9:54
quantilizers: ai that doesn't try too hard
-
18:05
intro to ai safety, remastered
-
5:51
why not just: raise ai like kids?
-
2:01:55
robert miles - "there is a good chance this kills everyone"
-
9:38
what can we do about reward hacking?: concrete problems in ai safety part 4
-
11:32
a.i learns to play tower defense
-
20:00
ai "stop button" problem - computerphile
-
11:44
ai learns to outrun police officers
-
6:56
reward hacking: concrete problems in ai safety part 3
-
11:47
we were right! real inner misalignment
-
11:32
how to keep improving when you're better than any teacher - iterated distillation and amplification
-
9:24
why does ai lie, and what can we do about it?
-
20:23
training ai to rob a bank
-
10:36
why would ai want to do bad things? instrumental convergence
-
36:02
chatgpt with rob miles - computerphile
-
9:40
9 examples of specification gaming