Four: Brian Christian on artificial intelligence
Effective Altruism: Ten Global Problems – 80000 Hours
00:00
Is It Possible to Flash That Out?
In the sing pages of your book, you say that humanity is kind of in danger of losing control over the course of events. But not so much to ae or machine learning specifically, but rather to models of the world in a more general sense. There's something dangerous, to my mind, about a model that's wrong, but has the ability to become right by a kind of terraforming reality to match the simplifications of the model. You can sort of think of climate change as this alignement problem where we didn't put into the reward function things like sea level and whatever.
Transcript
Play full episode
