“We’re All Gonna Die” with Eliezer Yudkowsky

Summary: We have no idea how to ensure alignment of a super AI. People are spending a lot of money to create a super AI. Super AI is likely to destroy humans as a byproduct of its actions, whatever it will decide to do. Happening to create a super AI that is also a humanity aligned super AI is like creating software randomly and it becoming information secure by accident, very unlikely.

Investors and researchers are mocking people warning of the large scale dangers of AI.

I really hope this gets circulation in HN. It's a very moving interview and I'm left with the impression that the best chance for a solution is a last effort in changing the minds of leading AI researchers to really engage with and take this problem seriously. I hope they will.