Chris Hayes Has a Warning About the Dangers of Trump’s AI Whims

Chris Hayes Has a Warning About the Dangers of Trump’s AI Whims

Late Night with Seth Meyers

0:01 -Our first guest tonight is a best selling author and the Emmy-winning host

0:03 of "All In With Chris Hayes," which airs weeknights at 8:00 on MS NOW.

0:07 He also hosts the podcast "Why Is This Happening?"

0:10 which just launched a new series, "The AI End Game:

0:13 Power, Profit, and Progress." Please welcome back

0:15 to the show our very good friend Chris Hayes!

0:18 Cheers and applause]♪♪ -Hello!

0:29 How are you?

0:29 -I'm great.

0:30 How are you?

0:31 -I'm a big fan of your podcast.

0:32 I'm very happy you're addressing the AI end game,

0:36 because a lot of people say that one of the end

0:38 games is just the total extinction of the human race.

0:41 -Yes.

0:42 And I don't know if you feel this way,

0:43 there's something about the tenor of the AI conversation

0:47 that makes me curl into like a fetal position.

0:50 -Yes.

0:51 -And it just gives me, like, the bad feeling.

0:53 -Well, even the people that are most excited about it often will then add,

0:57 like also a lot of people are going to lose their jobs.

0:58 -Yeah.

0:59 They're like, this is amazing.

1:00 And like, you're all going to lose your job and then

1:03 be subject to like some slave master HAL kind of thing.

1:06 And it's like, okay, great.

1:08 And they're like, but we have to build it.

1:09 -Yes.

1:10 -We just have to build it.

1:11 -I've literally heard people say like, it's amazing how fast you can,

1:14 like— it can write your resume like this.

1:16 I'm like, it doesn't— if there's no more jobs.

1:21 Like, what's the— great, I'm glad you saved all that time.

1:24 But like, so how, where are you landing on this?

1:27 -I think the reason that I undertook it was

1:29 precisely 'cause I was trying to get over that feeling,

1:31 that visceral feeling of defensive crouch.

1:33 -So you're trying to come out to like, face it with a little bit of optimism.

1:37 -To overcome that feeling, to feel like, okay,

1:39 if I start to actually get curious about this and navigate through,

1:43 I think the big thing I think is

1:46 in the sort of bell curve of possible distributions,

1:49 I think the most likely is that it is a revolutionary, but "normal" technology.

1:54 -Right.

1:54 -In the sense of like the car, the telegram,

1:57 the internet, the cell phone, the personal computer.

1:59 -And you've pointed out there were panics about all of these as well.

2:03 -Yes.

2:04 And also, all of those were transformative,

2:06 also did have tremendous negative consequences

2:09 and also arguably some amazing positive consequences.

2:12 But in the end, like, I guess the way I feel is we're still here,

2:16 we're still humans, we still interact with the world.

2:18 And I think the most likely outcome of the AI revolution,

2:22 if you want to call that, is the same— that we will be here.

2:25 We will be us.

2:26 -Yeah.

2:26 -We will have the challenges and problems we have,

2:28 and there will be new difficulties and challenges.

2:31 But understanding it in that context as opposed to nuclear winter.

2:36 -Right.

2:36 -I think is really important and useful.

2:38 And I think probably the most correct way to think of it.

2:40 The other thing I'd just say is if it is nuclear winter,

2:43 then like, you know, what are you going to do?

2:45 -Yeah.

2:45 -Right, like...

2:46 -You know what— -It's like...

2:47 if it's, if it's, if it's gonna be,

2:48 if it's gonna enslave all of us and we're all gonna be like in the matrix,

2:51 it's gonna be sucking our life force out for battery power...

2:55 -Yeah.

2:55 -You know, I guess that's our fate.

2:56 -Rock and roll, yeah.

2:57 -Exactly.

2:58 There's a question about regulation.

2:59 And, you know, the Trump administration was very hands off for a while.

3:03 Now they've sort of feinted towards the idea

3:05 that they might regulate it, but like,

3:07 do you have a lot of confidence in the way they would go about that?

3:11 Laughter] -I mean, I can't see a single problem with it.

3:15 -Yeah.

3:15 -I think— -Like they're the guys— -I like—

3:17 -They're the guys you want to be on this.

3:19 -I like the idea of Donald Trump personally in the models.

3:22 -Yeah.

3:23 -Like poring over the code, like with the frontier guys being like,

3:26 "Oh, what do you got going on here?" You know.

3:28 -Yeah.

3:29 -No, I mean, I think, you know, they have this totally hands-off view.

3:33 And again, you know, there's history here.

3:35 You know, when we started— when cars started racing around,

3:38 they started doing things like hitting kids in neighborhoods.

3:41 -Yeah.

3:41 -And people were like, what the heck is this?

3:43 You know, we had to come up with a thing called a speed limit.

3:45 We had to come up with regulation and imagine

3:48 the universe in which we were just like, We're just gonna let the cars...

3:51 -Right.

3:52 -Kind of make it up as they go along.

3:53 You know.

3:54 -Like it's the cars' world now.

3:56 -Exactly.

3:56 Like, no, obviously this is a powerful but dangerous

4:00 technology you have to establish some rules for.

4:02 The problem here is, of course,

4:04 their instincts are always sort of kingly or dictatorial.

4:07 So there should be some sort of comprehensive legislative agenda,

4:12 some kind of regulatory body, hopefully staffed with like technical experts,

4:17 which is an entire form of governmental institution they're trying to destroy.

4:22 -Right.

4:22 -Rather than, like, the whims of Donald Trump.

4:24 -Yes.

4:25 -But that's true for everything.

4:26 -This was something that did not inspire confidence in me.

4:30 In July, Trump compared AI to a beautiful baby that's born and said,

4:35 "We can't stop its growth with stupid rules." So I feel like that's...

4:41 Laughter] Like, this makes me think not only is he not an expert on AI,

4:46 but like also babies.

4:47 -I don't think— I don't think he's been spending a lot of time around babies.

4:50 I mean, you and I as both dads of three know.

4:52 -Yeah.

4:53 -The one thing you want to do with babies,

4:55 infants, toddlers is not give them rules.

4:56 -Yeah.

4:57 -It's the most important thing.

4:58 -Yeah.

4:58 You got to let them grow.

5:00 -And then they grow up and they get so big and so beautiful.

5:03 -Yeah.

5:03 -And, and they— and so— and they reach maturity in such

5:07 a great way if you just never tell them no.

5:09 -Yeah.

5:09 -Yeah.

5:10 No rules, no structure.

5:11 -That's exactly right.

5:12 -You just let them grow.

5:13 Ideally, you know, you just, you're not even around for most of it.

5:16 Laughter] You meet up with them again when they're 18.

5:19 -Yes.

5:20 Um, you know, I should— you know, all of this is true.

5:23 These are fair criticisms about how

5:24 the Trump administration would go about regulating AI.

5:26 I would also not feel a ton of confidence if,

5:29 like, Chuck Schumer was in charge of this.

5:31 -Fair.

5:31 -There's like, there's nobody on the Democratic side where I'm like,

5:34 those are the guys that could absolutely sort this out for us.

5:37 -No.

5:37 And I think if you go back and you look, again,

5:39 if you look at like the Telecommunications Act in 1996, I think it was,

5:42 there's all sorts of real critiques for that and the idea of some big

5:46 legislative apparatus to regulate this, there's going

5:49 to be a huge amount of industry capture.

5:50 There's going to be lobbyists writing it.

5:51 But again, it kind of comes back to that kind of, you know,

5:54 democracy being the worst system except for all the alternatives.

5:58 -Right.

5:58 -That's sort of where I am on this.

6:00 It's like, yes, do I feel amazing

6:02 about Congress writing some AI regulatory bill?

6:04 No.

6:05 Do I think it's awesome to just have the Frontier Labs themselves?

6:08 I mean, this happened with Anthropic recently.

6:10 -Yeah.

6:10 -They develop a new model and they're like, Oh, hey, look at this.

6:13 This model can find every zero day exploit,

6:15 every like massive vulnerability in every critical system around the world.

6:19 We're not gonna release it to the public.

6:21 Now, that was probably the right call.

6:23 -Yeah.

6:23 -It's a little weird that they're just

6:25 voluntarily sitting there with the nuke button.

6:27 They're like, we're not gonna touch it this time.

6:28 -Yeah.

6:29 -Well, you know, someone like that, that decision has to be

6:33 made under some form of democratic regulatory and legal supervision.

6:38 And yes, you're right.

6:38 You're right to be skeptical of anyone having the wherewithal to deal with this.

6:43 But in some ways, it's not like we haven't figured this out before.

6:47 -Right.

6:47 There is also, I mean, you know,

6:49 I think a lot of times this happens in politics, right,

6:51 where it's almost like a roulette wheel of when will

6:54 this hit this critical moment where Americans panic about AI, right?

6:59 Like, does it lead to massive job loss?

7:02 And ultimately, like, it's going to be a little random,

7:04 whether a Republican or a Democrat is in power when this moment comes, right?

7:07 -Yes.

7:07 -And, uh, you know, do you feel like

7:11 anyone is prepared for how, you know, again,

7:13 if it ends on the other side of this bell curve where

7:15 jobs just are leaving industries that have

7:18 depended on them for human employment?

7:19 -Yeah.

7:20 I mean, there's two, two things happening right now.

7:22 At the grassroots level, you are seeing across ideological spectrum,

7:26 across the partisan divide,

7:28 and across geographic areas, rebellion against data centers.

7:31 And this is happening in small, rural Republican areas.

7:35 It's happening in suburban Democratic areas.

7:37 People are like, we don't want these things.

7:38 -Yeah.

7:39 There's growing fear and sort of backlash at the ground level.

7:42 And then at the sort of higher level,

7:44 I think Bernie Sanders has been doing some interesting work on this.

7:47 I mean, one way to think about the project here is to do

7:50 to white collar workers what globalization and neoliberalism

7:53 and trade did to blue collar workers.

7:56 And it's like, if you think American politics are crazy now,

7:59 turn Marin County into Gary, Indiana, and Park Slope into Youngstown.

8:03 -Yeah.

8:03 -And then see how crazy it gets.

8:05 -Yeah.

8:05 -Due to white collar workers,

8:08 the people that have laptop jobs that feel like in a very unequal America,

8:13 in a very challenging environment,

8:14 they've managed to go to school and get a degree and have a job that could

8:18 provide the kinds of things that used

8:20 to be what the middle class could get, right?

8:22 And now take that away and say

8:25 you're now the factory worker who's been deskilled.

8:29 And what do you think's going to happen to American politics?

8:31 It's not gonna be pretty.

8:32 -I mean, the one thing that might unify everybody if

8:35 it's just like us against the robots, I feel like that.

8:37 -No, I mean, that— absolutely.

8:38 And I think, you know, Ron DeSantis in Florida the other day made, you know,

8:42 said something about this about, you know,

8:44 he's a relatively canny politician in certain ways.

8:47 He's got his ear to the ground about

8:49 this sort of brewing backlash at the local level.

8:50 So, you know, people in AI,

8:53 people at the Frontier Labs, they're saying this, too.

8:56 I mean, one of the things that they keep saying

8:57 is they do have this vision of massive job loss.

9:00 I mean, in some ways that's the value proposition.

9:02 -Right.

9:02 -The whole idea is you can pay this software engineer $200,000 a year,

9:07 or you could pay, you know, our program $100,000.

9:11 Get rid of him.

9:12 Save 100,000 and you keep the, you keep the surplus, right?

9:15 From their view, they think that's coming, too.

9:17 And they're also kind of weirdly panicked about it for that reason.

9:21 -Yeah.

9:21 Well, I think when they're like, everybody's going to remember,

9:23 they're like, hey, who do you work for?

9:24 -Right.

9:25 Yes, exactly.

9:26 -Yeah.

9:26 It's gonna be, it's gonna be trippy.

9:28 You have an interesting theory as to why Donald Trump does not seem dead

9:32 set on fixing the problems that are ailing America in regards to the, you know,

9:35 the economy and rising gas prices.

9:37 Your theory is he's bored.

9:39 -He's so bored.

9:40 -Yeah.

9:40 -It's— he's the— He's the most bored man alive.

9:44 -Yeah.

9:45 -He's the most bored dude in the history of the universe.

9:49 Like when you watch him at these White House events,

9:51 like everyone makes a big deal about him falling asleep,

9:54 but it's actually a state of waking boredom.

9:57 -Yeah.

9:57 -Where the eyes go down, it's just like, oh, I can't believe...

10:02 Is this event still going on?

10:04 -Yeah.

10:04 -And then every time— there's one thing that holds his interest,

10:07 there's one thing he's so interested in, he just wants to talk about,

10:10 which is the ballroom.

10:11 -Yeah.

10:11 -So any time it's like people ask him questions about like,

10:13 well, you know, no oil is getting through the Strait of Hormuz.

10:17 He's like, let's talk about the ballroom.

10:18 -Yeah.

10:19 -Have you heard about my ballroom?

10:20 -He's like, there's going to be a window there.

10:21 -Yeah, right there.

10:22 -Yeah.

10:23 -Maybe we can watch the Strait of Hormuz from here.

10:24 -It is like...

10:25 Laughter] And so he just, it is it's about like building arches and just like,

10:30 you know, the reflecting pool.

10:33 He just is like fully becoming a construction guy again.

10:36 -He's— that's exactly it.

10:38 He's, he's coming home.

10:39 He's coming home to himself.

10:40 He's a person who's obsessed with his legacy right now.

10:42 I think he's not a person— I mean, none of us,

10:45 I think, are great at facing our own mortality.

10:47 -Yeah.

10:48 -I would say he's particularly probably not great at it.

10:50 At all.

10:51 -Yeah.

10:52 -I don't think he's done a lot of work on himself.

10:54 I would say in therapy language.

10:56 -Well it's certainly like if he was in therapy and he's

10:58 like,[ As Trump] "I think I know how to fix it.

11:00 I'm gonna make a patio." -Yes, exactly.

11:03 -And the therapist is like, "I don't think that's gonna fix it." And he's like,

11:05 "Wait 'til you see the patio."[ Laughter] But it is,

11:08 I mean, it is kind of insane.

11:09 Like, I mean, the ballroom does feel like something

11:11 you'd really want to dig into at the therapy level.

11:14 -Yes.

11:15 And, and so much now, I mean,

11:17 you can't keep him off the ballroom for three or four minutes in a row.

11:21 -Yeah.

11:22 -I mean, he'll bring it back to the ballroom.

11:23 -Yeah.

11:24 -That's all that's occupying him.

11:25 And I do think it's like, it's become this kind of, yeah,

11:29 sort of psychological safe space for him to go back to and physical safe space.

11:33 -Yeah.

11:34 -That's the funny thing about it too, right?

11:35 He keeps saying it's like it's the ballroom, but it's also the bunker.

11:38 -Yeah.

11:38 -Ballroom on top.

11:39 Gilded ballroom on top.

11:41 Bunker below.

11:41 You know, where— where I'll be safe and away from all these people.

11:46 Like, what exactly are you planning, buddy?

11:49 -Uh, well, it seems like everything's in good hands.

11:52 Thanks for being here.

11:52 It's always a pleasure to talk to you.

11:54 You guys, this is Chris Hayes.

11:55 Cheers and applause] "All In with Chris

11:57 Hayes" airs weeknights at eight on MS NOW.

11:59 The AI End Game is available now, with new episodes out weekly across all

12:03 podcast platforms and on MS NOW's YouTube channel.

12:05 Right back with Meg Stalter.

Study with Looplines Download Captions Watch on YouTube