I cite your work on approval-directed agents as a promising direction for developing A.I.
Gabriel Murray

I agree that act-based agents in general can more easily incorporate ethical arguments that don’t fit into a consequentialist framework, whether virtue ethics or respect for rights or what have you. I think this is closely related to their robustness to current ethical uncertainty, and more generally to the kind of robustness that I think is necessary to actually work in the real world.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.