How About Tomorrow?
How Dangerous Can AI Get, Dax is Down on DeepSeak, and AI First App Development
Adam
So they have these tests, but they have to worry at level four that the AI will just like sandbag and pretend that it's not smart enough, even though it is because it knows it wants to pass the test. Yeah. Which is super interesting to think about. What do you do if these models can scale to super intelligence, smarter than us? How do you control something that's smarter than us?
0
💬
0
Comments
Log in to comment.
There are no comments yet.