#49 - AGI: Could The End Be Nigh? (With Rosie Campbell)
Increments
00:00
The Importance of Rewarding Systems
I think the argument is if you are repeatedly training these systems, then that will become a useful way to get reward. I wouldn't agree that computer viruses don't have the capability of thinking ahead. And second, I don't think reinforcement learning is necessarily more powerful than conventional ways of writing programs. It's just a tool that can be used in some circumstances,. Just like supervised learning can, just like unsupervised learning can and let's not forget just conventional programming.
Transcript
Play full episode
