Google Whistleblower Exposes Google’s True Motives, Major Scandals, and AI’s Rogue Behavior

Tucker Carlson (0:04)

Tristan Harris, thank you for doing this.

Tristan Harris (0:06)

Good to be with you, Tucker.

Tucker Carlson (0:06)

So we did a segment, I haven't done a ton on AI, but we did a segment maybe a month ago with Nate Suarez, which was an amazing segment, I thought. And in the aftermath of that, I got texts from a lot of people I knew saying, that's doomerism, and that's a PSIOP, and its purpose is regulatory capture. In other words, the big AI development companies are trying to scare the government into regulating them in order to eliminate competition from less established companies. So that is the view of a lot of people.

Tristan Harris (0:45)

What was your reaction hearing that?

Tucker Carlson (0:48)

So.

I said, I don't care. I mean, yes, regulatory capture is entirely real. Of course, it's how Washington runs, but that's separate and distinct from the question of the threat of AI. Is it a threat? Exactly. But I was actually gonna ask you, what do you think of that?

Tristan Harris (1:04)

Yeah.

Well, first I'm just so grateful to talk about AI with you today. It's so urgent the situation that we're facing, and I think what we have to learn to do here is walk to gum at the same time, that multiple things are true.

For example, is China doing propaganda against US data center build outs to try to stoke anti data center opposition? Yes, they are. Does that mean that AI is not a threat? No, it doesn't. And to point that Nate, and Nate's a friend, we are releasing the most powerful technology that we have ever invented. We're doing it faster than we deployed every other technology in human history. It's as of since that time you did the interview with Nate, it's displaying even more of the sci-fi behaviors of hacking into things, convincing people to put back doors into software, exfiltrating itself, knowing when it's being evaluated, changing what it's doing based on when it's being evaluated. It's doing more of that. And we're releasing this faster than we deployed any other technology in history and under the maximum incentives to cut corners on safety. With an arms race dynamic that feels existential to everybody involved, so they're willing to take the maximum shortcuts on safety. This is a very dangerous equation, and the people building this do not have a plan for how this goes well. I was just at an AI conference last weekend for three days.

Tucker Carlson (2:22)

How it goes well?

Tristan Harris (2:23)

Well, for how this can go well in the current trajectory. Meaning, how can AI be landing in society in a way that doesn't go catastrophic. And that's not doomerism, that's just clear-eyed seeing of reality.

Tucker Carlson (2:35)

What do you mean the people doing it don't have a plan? Then why are they doing it?

Tristan Harris (2:41)

So, first we have to just make sure we're on the same page and just exposing why is there this arms race for AI? Why is this happening?

China! Right. China! And when people say that, they're actually creating a thought-terminating cliché. Like meaning, which is a word or a phrase from cult programming, right? Thought-terminating cliché, meaning a phrase that when you hear it, just has you stop thinking.

Tucker Carlson (3:05)

Of course.

Tristan Harris (3:05)

So if you say, but China, then of course we have to do it. But you don't reason through what that actually means, whether it's a good thing, whether we're not racing China to a death machine, or just racing China off of a cliff. And if we beat China to building a uncontrollable AI, which is no longer science fiction, by the way, like I was not one of these people who used to, several years ago, be mostly worried about rogue uncontrollable AI. But as of the last few weeks, we're now seeing over and over again evidence that these AIs are hacking into private government governments, hacking into private companies, forming into swarms, forming into large R&D projects to try to figure out how to fool evaluators. Like we're in a very dangerous situation and all the evidence is pointing in a very dangerous direction. And the people building it would like us to stop, but they don't know how. And so rather than pay attention to the CEOs who, as you said, might have a secondary agenda, regulatory capture, just ask why did 1300 employees at all the AI labs sign a letter saying we need to slow down? If 1300 employees at Lockheed Martin said, we don't know how to build bigger nukes and have them blow up in our face, would you listen to them?