Non-Trivial AI
DarkNightoftheSoul
•
8mo ago
•
100%
Concrete Problems in AI Safety Part 2: Empowerment
https://www.youtube.com/watch?v=gPtsgTjyEj4Maybe AI systems would be safer if they avoid gaining too much control over their environment? How might that work?
Comments 0