Stuart Armstrong: Is AI an existential threat? We don't know, and we should work on it

On the 25 of November, 2020, Stuart Armstrong was part of a debate organised by York University, Toronto, on whether AI was an existential threat.

Instead of taking one side or another of the debate, Stuart argued that if you compare with similar technologies (the "outside view"), then AI seems safe. If you instead focus on the specific features of AI (the "inside view"), then AI seems potentially very dangerous. Despite this uncertainty, the risk is certainly high enough to justify work on preventing such existential risk.

This talk is an extended version of his presentation at the debate.

[diegoevrard-broquet]

best one so far! not too much background noise, voice doesn't sound asmry