Silicon Valley Rep: You CANNOT Trust Tech Billionaires On AI

Silicon Valley Representative Roan is introducing legislation that would put new restrictions on super intelligence, particularly banning recursive self-improvement. That's the ability of a uh of an AI basically to rewrite its own code to make it uh better and also implementing other uh safeguards. Representative Kana is joining us here to discuss. Thank you so much for being

here. >> Thank you. And so to sir, >> so uh Representative Connor just wanted to outline for us, you know, what this uh legislation does. >> Well, let's take a step back on the bigger picture. I represent Silicon Valley. Uh I know these tech guys. I've been in the rooms. You cannot trust tech

billionaires to write the rules to keep us safe. The idea that the president is saying that they can self-pol themselves is absurd. Uh we need regulations. Uh first we need to ban recursive self-improving AI. All that means is we need to ban AI that can train itself where human beings lose control. Second,

we need to have embedded in these companies uh auditors who can tell us what's going on. I'm all about enforcing liability laws and I agree with Lena Khan. enforce them and uh hold them civily and criminally liable. But we can't have a world where uh oops 100 million people die because of a

biological attack or AI loses control. Let's sue them. We actually need regulation to prevent that just like we have aviation and nuclear uh regulation. And third is we got to create that AI safety agency. And here's the most important point, Ryan. These tech leaders count on jargon to obfuscate. They use fancy words like sandbox, air

gap. They go to Washington and then people are, you know, stutter because they don't know the jargon. Uh there are plenty of really, really smart people uh in AI safety who don't make profits. They're all in this one building at Berkeley. They're the AI safety community. Start to appoint them into an independent agency that can regulate