Dylan Hadfield-Menell, UC Berkeley/MIT: The value alignment problem in AI
Generally Intelligent
00:00
The Incomplete Contracting Line of Work in Machine Learning
Incomplete contracting looks at the challenges that arise when you can't specify a perfect contract on performance incentives. This is where the terms need to be human interpretable. That's where having institutions that can define them and manage p r campaigns will help, he says.
Play episode from 01:16:23
Transcript
