AI X-Risk: Difference between revisions
| (2 intermediate revisions by the same user not shown) | |||
| Line 1: | Line 1: | ||
=Risk Spectrum= | =Risk Spectrum= | ||
= AI Risk Spectrum (Representative Thinkers) = | |||
{| class="wikitable" | |||
! Position | |||
! General View | |||
! Representative Thinkers | |||
|- | |||
| Strong extinction risk (“p(doom) high”) | |||
| Superintelligent AI is likely to cause human extinction or irreversible loss of control unless development is halted or radically redesigned. | |||
| Eliezer Yudkowsky, Connor Leahy | |||
|- | |||
| Serious risk but potentially solvable | |||
| Advanced AI presents major existential risks, but safety research, governance, and technical controls may mitigate them. | |||
| Geoffrey Hinton, Max Tegmark | |||
|- | |||
| Moderate concern | |||
| AI poses real risks that require oversight and safety research, but extinction scenarios are uncertain. | |||
| Yoshua Bengio | |||
|- | |||
| Skeptical of extinction framing | |||
| AI risks are manageable engineering challenges; concerns about superintelligence are overstated. | |||
| Yann LeCun, Andrew Ng | |||
|} | |||
=Key presentations:= | =Key presentations:= | ||
| Line 25: | Line 47: | ||
Max Tegmark - Life 3.0 https://www.youtube.com/watch?v=Gi8LUnhP5yU&t=13s | Max Tegmark - Life 3.0 https://www.youtube.com/watch?v=Gi8LUnhP5yU&t=13s | ||
Yoshua Bengio - https://www.youtube.com/watch?v=azOmzumh0vQ | |||
Yann LeCun - https://www.youtube.com/watch?v=5t1vTLU7s40 'doomers think people are fundamentally bad' | |||
Latest revision as of 04:50, 9 March 2026
Risk Spectrum
AI Risk Spectrum (Representative Thinkers)
| Position | General View | Representative Thinkers |
|---|---|---|
| Strong extinction risk (“p(doom) high”) | Superintelligent AI is likely to cause human extinction or irreversible loss of control unless development is halted or radically redesigned. | Eliezer Yudkowsky, Connor Leahy |
| Serious risk but potentially solvable | Advanced AI presents major existential risks, but safety research, governance, and technical controls may mitigate them. | Geoffrey Hinton, Max Tegmark |
| Moderate concern | AI poses real risks that require oversight and safety research, but extinction scenarios are uncertain. | Yoshua Bengio |
| Skeptical of extinction framing | AI risks are manageable engineering challenges; concerns about superintelligence are overstated. | Yann LeCun, Andrew Ng |
Key presentations:
Stuart Russell - Nobel Prize winner - balanced discussion - https://www.youtube.com/watch?app=desktop&v=P7Y-fynYsgE
American Compass - Nate Soares and Eliezer Yudkowsky - https://www.youtube.com/watch?v=O4XXkO3uo8c
Robinson's Podcast with Eliezer - https://www.youtube.com/watch?v=0QmDcQIvSDc&t=3s
Center for Humane Technology - Tristan Harris and Aza Raskin - https://www.youtube.com/watch?v=xoVJKj8lcNQ
Technical discussion - Lex Fridman with Nathan Lambert and Sebastian Raschka - https://www.youtube.com/watch?v=EV7WhVT270Q&t=5s
ASI with Nate Soares - https://www.youtube.com/watch?v=0tjOzQne1LY
Connor Leahy https://www.youtube.com/watch?v=7Y_1_RmCJmA
Max Tegmark - https://www.youtube.com/watch?v=OkG5S1NwwVM&t
Joscha Bach - how intelligence actually works - https://www.youtube.com/watch?v=3MkJEGE9GRY
OpenClaw - viral agent - https://www.youtube.com/watch?v=YFjfBk8HI5o
Max Tegmark - Life 3.0 https://www.youtube.com/watch?v=Gi8LUnhP5yU&t=13s
Yoshua Bengio - https://www.youtube.com/watch?v=azOmzumh0vQ
Yann LeCun - https://www.youtube.com/watch?v=5t1vTLU7s40 'doomers think people are fundamentally bad'