00:00:00
You're one of the three godfathers of AI, the most cited scientist on Google
00:00:05
Scholar, but I also read that you're an introvert. It begs the question, why
00:00:08
have you decided to step out of your introversion? >> Because I have something to say. I've
00:00:13
become more hopeful that there is a technical solution to build AI that will not harm people and could actually help
00:00:19
us. Now, how do we get there? Well, we'll go through that. >> But is it fair to say that you're one of
00:00:24
the reasons that this software exists >> amongst others? Yes. >> Do you have any regrets? Yes, I should
00:00:30
have seen this coming much earlier, but I didn't pay much attention to the potentially catastrophic risks. But my
00:00:37
turning point was when Chad GPT came and also with my grandson. I realized that it wasn't clear if he would have a life
00:00:44
20 years from now because we're starting to see AI systems that are resisting
00:00:48
being shut down. We've seen pretty serious cyber attacks and people becoming emotionally attached to their
00:00:54
chatbot with some tragic consequences. >> Presumably, they're just going to get
00:00:57
safer and safer, though. So the data shows that it's been in the other direction. It's showing bad behavior
00:01:02
that goes against our instructions. >> So of all the existential risks that sit
00:01:06
there before you on these cards. Is there one that you're most concerned about in the near term?
00:01:10
>> So there is a risk that doesn't get discussed enough and it could happen
00:01:14
pretty quickly and that is but let me throw a bit of optimism into all this because there are things that can be
00:01:21
done. >> So if you could speak to the top 10 CEOs of some of the biggest companies in
00:01:25
America, what would you say to them? So I have several things I can say