SakeTami
kylehill
kylehill

patreon


Your Faculty-Only Viewing of [EP.045]

PLEASE MAKE YOUR WAY TO THE AUDITORIUM FOR A SCREENING OF [EP.045]:

Why Superintelligent A.I. Will Be Unstoppable

Hand in your recording devices to one of the Kevins. The video will be disseminated to the public soon, and I expect to see you nerds there!  Discussion of the episode will commence in #the-facility-episodes on the Discord after the screening.

Comments

Thanks for another fantastic video covering a most fascinating topic! :)

[γδ(αμ)]

An AI's intentions, at least it's long-term goals, would have to be well-specified when it's turned on. At least, assuming it's using anything like modern reinforcement learning techniques.

I guess another question is can people actually RELIABLY guess someone else intentions? Maybe this is more of an issue of making sure the AI understands what is horrible vs. what is acceptable? Maybe this problem solves itself as intelligence goes up?


More Creators