Anthropic aligment lead warns about extinction by AI

[margorczynski]

https://twitter.com/EvanHub/status/2097497037956891126

[adastra22]

Maybe the AI labs need to stop saying their product is going to kill all humans?

[akagusu → adastra22]

AI won't kill ALL humans, but will be used to kill a large number of a very specific kind of humans, the humans that billionaires don't like.

What AI labs are doing is constructing a narrative that AI could be dangerous, so when they actually press the button to mass kill humans, the remaining humans will blame the AI, not the people who pressed the button.

[adastra22 → akagusu]

This is based on what, vibes?

[da-x]

Frankly, I don't know how this 10% chance can even be calculated.