Will the Tech Oligarchs Turn on Trump?

Subscribe here: Apple Podcasts | Spotify | YouTube

In this episode of The David Frum Show, The Atlantic’s David Frum opens with his thoughts on the growing calls for Medicare for All. David discusses how Medicare for All might actually work, why it’s impractical, and why it is fundamentally a promise Democrats cannot keep.

Then, David is joined by Kara Swisher to discuss her new podcast with Hillary Clinton, the dangers of unregulated AI, and the power of the AI oligarchs over American politics. David and Kara discuss the potential of an AI bubble and what responsible oversight of AI would look like.

Finally, David ends the episode with a discussion of The Rise and Fall of American Growth: The U.S. Standard of Living Since the Civil War by Robert Gordon, and whether the AI revolution changes any of Gordon’s pessimistic assessments of what has happened to American growth since 1970.

The following is a transcript of the episode:

David Frum: Let me ask you about what works and things you’ve changed your mind about.

So you have been a career-long advocate for stronger regulation of technology in general—

Kara Swisher: David, let me correct you. Any regulation. We don’t have any.

Frum: What’s your line? “Your local deli is more regulated than the biggest tech platforms in America.”

Swisher: Everything.

[Music]

Frum: Hello, and welcome to The David Frum Show. I’m David Frum a staff writer at The Atlantic. My guest this week will be Kara Swisher, host of the On With Kara Swisher podcast, and co-host of Pivot, and co-host, very excitingly, of a new podcast that she will co-star [with] Hillary Clinton. And we’ll be talking about that new podcast with Kara.

This week’s book is The Rise and Fall of American Growth: The U.S. Standard of Living Since the Civil War by Robert Gordon, an enormously influential book on economic history published a decade ago. And I’ll be discussing whether the advent of the AI revolution changes any of Robert Gordon’s pessimistic assessments of what has happened to American growth since 1970.

But before the dialogue with Kara Swisher and before the book discussion, I want to open with some thoughts about the debate over Medicare for All: the slogan, and the idea of Medicare for All.

It does look like the next Congress will feature many more advocates of the Medicare for All idea than American politics has seen for a long time. There are bills in Congress purporting to offer or designing or proposing a Medicare for All scheme. It’s a slogan on the campaign trail. Let’s take it seriously. Let’s think about this in a real-world way, because we may be talking about this a lot come 2027. What does Medicare for All mean? How would you get there if you were minded to get there?

So understand, Medicare, of course, is the program that provides much health insurance for people over 65. It has components with prescription drugs, but the core of it is that it covers medical and hospital bills for people over 65, with some contribution from the over-65s themselves.

The Medicare for All proposal is to extend this scheme through all of the American economy, covering everybody through the Medicare program and displacing a lot of the health-care insurance that you have now. Which is mostly—those people who get private health insurance get it usually paid for through an employer, out of the worker’s pay, but through an employer.

What would it mean to make this change from a system where much of the health care in the economy is paid for by employer-financed private insurance to one where the U.S. Treasury pays for most health care? So: Half a decade ago, at the end of the 2010s, there was a big study of this. The Congressional Budget Office, and others, too, did a big study of what it would mean to transition to a Medicare for All program. And CBO’s estimate in the year 2020 was that such a program would cost the U.S. Treasury, over the next 10 years, about $30 trillion. The work has not been redone by CBO. But if I think it—if you just figure out prices are higher, you’re going further out into time, you’re starting in 2030 and going to 2040—it’s a good guess that it would cost closer to $40 trillion over 10 years than it would cost to $30 trillion over 10 years.

So where would you get the $40 trillion? The U.S. Treasury is going to be spending a lot more. How would it get the extra $40 trillion? Now, proponents of Medicare for All will say—and they’re not completely wrong about this—they will say: Wait a moment. That’s not $40 trillion of new spending. We’re spending the $40 trillion now or thereabouts. We’re spending it one way. That is, you get mostly from your private insurance through your employer. So we’re just taking $40 trillion from this sector of the economy and moving it over to the government, and net-net, it will even out. Maybe it’s a little bit more expensive than what we have now, but not a lot more expensive.

Let’s suppose they’re right about that. Still, we have one way of financing it now, and we’re going to be moved to another way of financing it along the way. How do you make that transition? What would that look like? The proponents of Medicare for All will say this: Look, when you get a job, if you have health care, you are really paying for the health insurance. Your employer lowers your wages to cover the cost and then pays an insurance company, or they self-insure. But in any way, the employer finances the payment of the health coverage out of the wages that you don’t get. When we move to health care for all, Medicare for All, the employer will no longer be buying private insurance. So your wages should go up, and out of that higher wage, you should be able to pay a higher tax.

But that blackboard analysis blinks the fact, yeah, you’re going to be paying a new tax, probably a payroll tax, maybe some other kind of tax. And it’s going to be quite a big tax that is going to fall on just about every worker in the economy.

Now, again: You’re paying a kind of health-care tax now, because your wages are lower. But the Medicare for All people assume that your employer will raise your wage almost all of what you’re now paying for health insurance through your employer. You’ll get almost all of that money back, maybe all of it. That’s the assumption. And so you’ll be well-positioned to pay the new tax, which will not be heavier than what you’re paying now. But even if that’s all true, it won’t be true for everybody. Some people will be winners; some people will be losers. Some people will be big winners, some people will be big losers, and everybody will see a new tax on their paycheck.

So what Medicare for All means from the point of view of American politics is: The new president elected in 2028—assuming that president is committed to this idea—the new Congress that comes to power either now or in 2029, that new Congress, one of their first orders of business will be passing the biggest and most far-reaching tax increase that Americans have seen on their paychecks since World War II. That’s what Medicare for All means. Start the next Congress, start the next presidency with a massive increase in a visible tax that you explain to American people they won’t really feel—because their wages should be going up if their employer gives them back the money the employer is keeping right now to pay for health insurance.

And aware of the politics of this, some of the Medicare for All proponents have come up with a series of gimmicks to conceal the tax. In 2019, Senator Elizabeth Warren proposed a system where we wouldn’t have a payroll tax. We’d tax large employers in the economy directly. Every employer with 50 employees or more would pay a new tax directly on the employer. It would be invisible to the employee, and that would get us to the transition to Medicare for All.

But the Warren idea has been almost universally derided by experts in the field as completely complicated, unworkable, inviting all kinds of distortion. For one thing, it would mean that no employer in the economy who had 49 employees would ever hire a 50th. The 50th employer would generate a massive new tax on that employer. So on the verge of this AI revolution that is calling into question all kinds of people’s employment, you’d create a massive new tax that would say, If you’re at 49, stay there forever. Never get to 50 employees. So that seems kind of daunting and distorting.

So even if you believe all of these things, you’re signing up for a big project. Now, the really irresponsible people in the Democratic Party, the Bernie Sanderses, will say, We don’t need payroll tax. The billionaires and the millionaires: They’ll pay for it all. But that’s, again, more sloganeering. The math doesn’t work, and anyway, billionaires and millionaires find it quite easy to escape taxes. And the way you would write the tax would be with an increase in income taxes—and that would be seen not just by billionaires and millionaires, but by lots and lots and lots of people.

There is no way around it. Medicare for All means transition to a new kind of program that needs a new kind of funding mechanism that requires a new and highly visible tax that will fall on many, many people. And if you think no Congress will ever pass that thing—which is kind of what I suspect—then when you talk about Medicare for All, you are promising something that you know you cannot deliver.

One of the reasons I think that American politics has seen so much demoralization and polarization over the past 15 years is because you’ve had politicians—President Trump is the worst offender, but he’s not the only offender—who promise things they know can’t happen. You’re not going to restore American manufacturing and make America great again by keeping out foreign products. You can’t pass tax cuts forever and have no spending cuts. Those things, it doesn’t work. Gravity, math—they all exist. And people, if not the candidates themselves, people around the candidates, they know they’re promising things they can’t do and Medicare for All is just such a thing.

Now, where Medicare for All is very useful for a politician, and this goes back to the Bernie Sanders model, if you’re the one senator who is in favor of this idea, and all the other senators say, It’s unworkable. We’re not going to do it, then you can get enormous attention for yourself. You can pose as the one truth-teller in the Senate, the one honest man in the village of the accomplices and the weaklings. You’re the one guy, and you’re saying this one thing, and you don’t have to worry about the unfeasibility of it because no one will ever let you do it. And that’s how you get the career of Bernie Sanders, which is a career of decades in public life, almost no accomplishments of any kind, but an enormous following because of the promises he keeps making.

But you can only have one of those guys: one in the Senate, maybe a handful in the House. If you have a dozen senators who are promising things like that—if you have 40 or 50 or 60 or 80 or 100 members of the House who are promising those things—the question of the design of the program, the question of the how this is going to be paid for, the question of the massive tax that will fall on Americans on the promise that their wages will go up because the employers will give them back the money the employers are now paying for private health insurance, all of that really comes to the fore.

One of the ways to get to a better politics in America is for politicians to make promises that they know they can keep and that they intend to keep. And one of the ways to keep us on the bad path we’re on now is to have lots of people making promises that sound good to a comparatively small coterie of the voters, but that they know or should know they can’t keep. That they know or should know that they won’t keep. And that they can know or should know are impossible.

We’ve got a lot of that going on right now, and it points the way to further demoralization, further polarization. Change in an advanced society like the United States, in a complex bureaucratic society with a lot of vested interests—it has to be incremental. Even if incremental change is imperfect, that’s the way it has to be, and anyone who promises you a non-incremental change is promising something they can’t deliver and probably don’t mean.

And now my dialogue with Kara Swisher.

[Music]

Frum: Kara Swisher hosts On With Kara Swisher, co-hosts Pivot with Scott Galloway, and has a CNN series on the people trying to live forever. She has just announced a new election-season podcast with a co-host you may know, Hillary Clinton. Kara Swisher is known to her many followers as a tireless and fearless reporter on technology and those enriched by technology.

She is the author of three books, most recently Burn Book: A Tech Love Story from Simon & Schuster, in which she calls Mark Zuckerberg the most damaging man in tech and Elon Musk the most disappointing.

On Election Day, she’ll be onstage in Toronto for a live taping of Pivot, so Canadian viewers and listeners may want to make a note of that.

Kara, welcome to the program. Thanks so much for joining me today.

Swisher: Thank you. Unfortunately, we’re sold out in Toronto because the Canadians love us.

Frum: Oh, congratulations. Congratulations.

Swisher: Thanks, thanks.

Frum: Getting Canadians out on a dark November night.

Swisher: Well, we’re going to get them warm. We’re going to warm them up with dick jokes and insights, so that’s what we’re going to do.

Frum: All right. So let me start by asking you about your new co-host. How did that come to be? Tell me about the project.

Swisher: Oh, you know, it’s really interesting. One of the things I do a lot, and I have done over my career, is I kind of zig when others are zagging or I do something unusual. And, you know, there’s this sort of trope that all Democratic-establishment older people don’t know what they’re talking about, and they need to move aside. And I have said, many of them, I have said directly they should do that. But when I was at this dinner party—and I’ve known Hillary for a long time, and I’ve interviewed her many times on different things I’ve done over the many years—and it was at this, one of these big fancy tables. And Hillary had me sit next to her because we have a really good rapport, and we sort of tease each other a lot in really, in ways people don’t, I think, tease her.

And she had a mind full of really great, fast insights that were fresh to me that I hadn’t heard. And I hear all the insights. And then, you know, the economy: insight. Mamdani: insights, and actually in a way that you were not expecting from her. I was like, it was quite smart, very clear. And I thought, What is in her Wheaties these days? Because she really—she didn’t care anymore, and of course, this is what you hear off the record. And I thought: This is the Hillary you see sort of in those Lauren Boebert/Epstein hearings or in the Benghazi hearings. Very smart, very sharp, often on social media. And I thought: I’m going to bottle that Hillary.

And I just knew. And by the way, it’s racing up the charts right now. We’re now at No. 4, which is, I know that it’s because of her. And it’s because she has these fresh insights, and I wanted to show people the Hillary I know on the topics she’s good at. And so it’s worked. And we have—the rapport is important. But anyway, that’s what we’re doing, and we’re trying it out. And if it really works as both a business and a piece of content, we’re going to keep doing it through the presidential.

Frum: So is it the two of you ping-ponging with each other, or will there be third parties there? How will this work?

Swisher: There’s going to be guests. We just did one to start with. Well, I did it very quick. I do everything I do very quickly, essentially. And it was a germ of an idea, but it’s like Hillary unplugged. Or like, you know, when Johnny Cash came back and was doing Nine Inch Nails? That’s what I feel like I’m doing here. You do want to hear her. And I had tested it on Pivot, where she was spectacular. And we’ll have guests. I want people who disagree with her and who she disagrees with, both on the left and the right. Because that’s, to me, when she’s at her best. And really have some throwdowns. So people—I’m talking about with reasonable people, not insane people. I’m not going to have Laura Loomer go against Hillary Clinton, because it’s kind of ridiculous and pointless. But I want to have robust discussions about difficult issues, and I think, by smart people with substance.

Frum: Will you be able to get her to defend the person she actually is? And what I mean by that is: those people who were prominent in the Democratic Party from 1992 to 2008-ish, 2010. The people who believed in making government more efficient, in expanding trade, and bringing Mexico into North American economic life. That’s what she was associated with, and they came under attack from their party’s left. And they seem just to be, like, shamefacedly mumbling an, Oh, you’re so right, and The establishment was so bad, and Yes, we should burn everything down. But will she, on your program, be willing to stand up for the person she was?

Swisher: Yeah, I think she’ll defend it, and also say what’s good and bad. What I was really struck with on the Pivot podcast is, I was expecting her to do the sort of typical Democratic-establishment Mamdani thing. And she had a very insightful reason why he’s doing well. Here’s where he could go wrong, and stuff like that.

You know, look: Fact of the matter is, she’s been a, whatever you say about the presidential race. And I thought that there’s all manner of complexity there. It’s not just because Hillary lost; Donald Trump really won that election. And the changing tide of the American public, and the sort of attacks that the right had on her for years that end up working after a while. I think she will be very honest. I think she’s at a point—and she did say this—is, people like her more when she’s not running for anything. They like her more when she’s either in office or in service of a, you know, typically a man. It’s a little bit, you know, if I lived back then, I would’ve liked to have Eleanor Roosevelt do a podcast with me. Like all the insight she gained over the years.

And a good thing that she did this week: We were talking about the execution, which has now been canceled publicly by Donald Trump. She said: He’s going to cancel it, and here’s why. And she was 100 percent right. And she analyzed it, and then threw in what they did during the [Osama] bin Laden raid. Which was very helpful to understand the complexity of, like—there were people in the room that wanted to show his bullet-riddled body. They brought in experts; they brought in all kinds of things saying what the best outcome would be. And so she’s got a lot of wisdom sort of hidden in there that can be relevant to today. And again, she’s going to have to admit mistakes. She knows that, and she’s actually willing to do it.

Frum: I’m sort of more interested to hear her not admit mistakes if she doesn’t believe she made them.

What I mean by that is: I’m a graduate of a Republican world where the rule of the game was, if you were in the center of the party, you had a big kick me sign on your back. And the right of the party could say anything about you, and you could never say anything back. And the Democrats seem to have that same relationship with their left.

And I’m thinking about, so Hillary Clinton was instrumental in the negotiation of the Trans-Pacific Partnership in the Obama administration. She was proud of it. She believed in it. She was right. She defended it in the hardcover edition of the book she published before the 2016 election, and she stripped it out of the paperback edition. And you say, Why not stand up for what you believe? Why the kick me sign?

Swisher: She might have changed her mind, David. It’s called changing. She may have thought more about it and think about it, and—one of the things I really like about her is she thinks in real time. Like, she’s really compared to a lot of Democratic people who would just spout the same crap, essentially, like Everything’s fine. I just feel like she’s willing to throw down, all kinds of ways.

She’s also willing to be flexible about good ideas. And I think where she is right now in her career—she should speak to you about this—but I think where she is in, what fucking works, right? Like she started talking about the execution like, Of course it’s heinous; of course it’s amoral. But let’s not focus on that. Let’s focus on: If they do this, it’s going to be a recruiting video for jihadists. So let’s not do that. She’s very practical. And so I think she will—”whatever works” seems to be where she is in her life. And she’s not worried about, like, The Democratic establishment shouldn’t speak. She doesn’t care. She doesn’t care, and she shouldn’t.

Frum: Let me ask you about what works and things you’ve changed your mind about.

So you have been a career-long advocate for stronger regulation of technology in general—

Swisher: David, let me correct you. Any regulation. We don’t have any.

Frum: Okay.

Swisher: So not stronger.

Frum: What’s your line? “Your local deli is more regulated than the biggest tech platforms in America.”

Okay. So some of the AI executives have, in recent months and years, begun to repeat what you have been saying earlier. And my question to you is, when you hear that, do you think, Okay, look, they’re in a death race, and it drives everyone to do things they don’t wish to do. And if only the competition could be a little less fierce, then maybe they could think more about safety? Or do you think there’s something nefarious going on, that they’ve got some other agenda? Do you take them at their word when they say these things? Are they reformed characters? How do you read it?

Swisher: No. I mean, my approach to most of these people—not all of them, most of them—is, Hello, he lied. Essentially.

And so I think they—it’s a little complex. You know, it’s more complex than the way we binary every single thing we do around technology these days. They’re either villains or they’re heroes. Or they’re this, or they go from this. Or we love them, we hate them. That kind of thing. And life’s not like that, essentially.

What I think is happening is: They hate regulation. They don’t want any. That’s why they sucked up to Trump the first term, and that’s why they sucked up to him this time. Some of them had had real political conversions, but for the most part, they want to make money. They want to dominate. And that has never changed in all the time I’ve known them.

So, as I joked, if Kamala Harris had won, Mark Zuckerberg would’ve identified as they/them if it suited him. They’ll do whatever it takes. And by the way, guess what’s going to happen when the Democrats come in? Oh! He was the worst. I can’t believe I went to dinner at his house 412 times in the last—

Frum: Thank God you got here.

Swisher: Oh!

Frum: I’ve been waiting in this basement, waiting for you.

Swisher: I wasn’t really having a good time at that lunch.

These people literally would sell their mamas down, if they needed to. That’s my feeling on the general populace. And they’re very interested in winning. Winning and shareholder value. I mean, if Tim Cook is going to give a golden statue to Donald Trump, you’ve got to know what’s going on there. Now, I’ve heard since then he actually likes Donald Trump, which—okay, fine. No accounting for taste. But I think it’s all about shareholder value, every single thing they do.

And so I think when they talk about regulation, they want the regulation they want that is good for them. They never think of safety unless they’re forced to. But neither did big pharma, neither did airplanes, [the] car [industry]. Hello, Unsafe at Any Speed. That’s what it took. They wouldn’t have put seat belts in there if they didn’t have to. Or airbags. The chemical companies would love to spew poison, and, Oh well; some people die. And those people were civically interested. The tech people have never been civically interested, ever. Like, they don’t care about other people. And so we just have to make them have safety. And should, because this is very much like nuclear proliferation. This is on the same level of importance. Why are we outsourcing our safety to private industry on a topic that they themselves say is existential?

Frum: Well, this is the thing I find sort of baffling. There are people who say, and there are AI skeptics who say, You know, it’s not that big a deal. It’s the latest computer thing. It’s not transformative. There are people who say that. And if that’s what you believe, okay, yeah, have a light footprint, because how heavily do we regulate word-processing technology? And then there are people who say it’s going to kill us all. Well, if it’s going to kill us all, then it should be illegal. Government should not just regulating; government should be suppressing it. I don’t understand the people who take the view: I want to run a big company that makes products that I fear are going to kill us. Why are you doing that?

Swisher: I think the “kill us all” thing is—you know, with the way tech people talk, you’ve got to have listened to them for so long. They love to say “non-zero chance.” I was doing an interview with Elon Musk once, and he’s like, There’s a non-zero chance we’re in a simulation. Meaning we’re basically a video game for a greater civilization and teenagers are playing Earth, whatever it happens to be, and we aren’t real. We’re all a simulation. Well, it is a non-zero. It’s like there’s a non-zero chance that there is a god sitting up on a cloud. And so that’s what they tend to do.

And so what they know is—I think Elon actually said it best. It’s like, this isn’t just nuclear bombs; it’s things that can set off nuclear bombs without you wanting it to, essentially. And I often recount this dinner party I had in Silicon Valley 10 years ago. Many of the big names right now, they were posing these questions: What would, if you told AI to solve world hunger, what would it do? It would kill half the human race, because that’s the right answer. It’s not the right answer. It’s the answer that would work the quickest, right? And so they just—I think there is an existential crisis. But they don’t—

Frum: Wait a minute. Let me stop you on that. That’s a terrible answer actually, because—even if you’re a Nazi—because killing half the human race is a lot more expensive and time-consuming than feeding the 10 percent of the human race. Do you—

Swisher: Oh, they would do it with a nuclear bomb, my friend.

Frum: Do you think this is a line of work that attracts people with severe personality disorders? And that that distorts their thinking about problems?

Swisher: No, I’m thinking the AI would do it, because it doesn’t have any emotion. It isn’t attached to humanity. When Elon started off, it was interesting. I’ll use him because he was the earliest person to talk to me about the threats—he and [Open AI CEO] Sam [Altman], but mostly Elon—about the threats of AI, and was very concerned. And I think, you know, he’s a science-fiction guy. It also informs him. A lot of these people are science-fiction types. This is part of the air they breathe, essentially. And his first thing was Terminator. And we all have seen that movie where, you know—time machine, comes back, kills the original, blah, blah, blah. And so that’s where they were in sort of this dystopian idea. And that gives the machines a personality, an anthropomorphic personality, when in fact it doesn’t. Or like The Matrix, like they want to keep us as batteries, or they have a malevolence. It’s always a malevolence. Mostly it doesn’t have malevolence at all, because it doesn’t care. And doesn’t care—I don’t mean “hates us.” It means we don’t exist.

So he moved on to the idea that AI treats humans like house cats. Like, we don’t want to kill our house cat. We’ll feed our house cat. But that’s how they look at us. Again, a little too anthropomorphic.

The third one I thought he came upon was the best one. This was later. Which is: We’re building a highway in Georgia, and we put down asphalt. Do we know we’ve killed a lot of ants and anthills along the way? Do we think about it at all? No.

Frum: Okay. So what I meant by that crack about personality disorders is: I wonder whether when we inevitably anthropomorphize everything; that’s the way our minds are constituted. If you have a severely dysfunctional personality, when you anthropomorphize, you imagine the AI to be like you. So when Elon Musk thinks What would an AI be like?, he thinks it would be like him. And, yeah, kind of indifferent to human suffering. I wonder if a more normal person who had read, say, Frankenstein instead of The Terminator might say that this may be a more complex, tragic, meaningful relationship. And the limited—the kind of odd, odd stuntedness in some vectors, combined with obviously great talents in others—makes people poor understanders of the world around them.

Swisher: Yeah; they like to understand everything. The machines have power when they have power. They like to offput their power and their blame onto the machines. And it’s not the machines that are the problem. It’s like saying electricity is the reason we electrocute people. Or the knife was the reason people died. It’s because the person with their hand on the knife killed someone. And so they just want to blame the knife, right? And actually, recently, Jeff Bezos talked about this, like: We can’t be against the knife. We’re against murder. I’m like: Yeah, we’re against murder by you. Like, they don’t like to take that step by focusing in on the machine.

And one of the things they tend to do, a lot of, is remove themselves from the equation—that they aren’t at fault in any way. And this is a longtime thing, you know, with social media. Oh, I don’t know why girls are feeling self-esteem problems. Do boys listen to more porn? I wonder what caused that. Like, they’re very much not of that. And one of the things they’ve done, which I have followed of theirs, is just that, one: that they’re the victims. And two: that they have no consequences for the things they make. And this is the same kind of thing. This is what they made. And my only argument is like, why are we letting them decide on something this important? Because they’re broken people, and we need a bigger pool of people making a decision about something that they themselves say is very important.

Frum: Well, let me ask you about another one of your recommendations that has come true. You argued often that the ownership of TikTok should be shifted from its Chinese founders. Congress passed a law calling for just that. The Supreme Court said that that law was constitutional. Congress had the power to force the sale of TikTok, at least its U.S. operations, to a U.S. operator, and it eventually did happen. But what seems to have happened is, none of the things that are toxic about TikTok have changed. It’s just TikTok has been sold in ways to enrich President Trump’s allies and supporters.

When you look back on that, do you think, Hmm, that’s not what I meant. They didn’t hear me right?

Swisher: I was worried at the time. Many people were worried about the deleterious effects of social media, which are very real, in lots of personality, etcetera. I was worried about propaganda. I was a propaganda-studies major at the School of Foreign Service. So I think about—what could you use these things for? And if I were the Chinese Communist Party, that’s exactly what I’d do. I mean, I’m not mad at them for doing it. It’s just, that’s what you would do to have this. And we had no such service in China to be able to do that. I mean, our best thing is to lob Voice of America over there, and it doesn’t really work. The internet possibly is a way to get through into that culture, but they strictly control their information diet there.

And so I was worried about propaganda. Then I was worried about the ability to shift the narrative with the algorithm. But again, that’s gone back to William Randolph Hearst. I mean, this is not new.

Frum: Where I’m pressing on this is—the Trump administration seems like the nightmare version of the advice you’ve been giving. They have an activist Federal Communications Commission, which is on the march in all kinds of ways to stop things that they don’t like, whether it’s The View or Jimmy Kimmel.

Do any of your feelings about regulation change watching the Trump administration wield these regulatory powers?

Swisher: No, because—we have got to go back. Obviously, after post-Nixon, they went back and fixed a lot of things they didn’t think he would do. We’re going to have to go back and figure out how to deal. If the internet isn’t regulated by the government, then neither should broadcast networks be. It’s not a public good. I mean, at this point, we should just sell it to them, and they should do what they want, and there probably should be some sort of public-access kind of thing.

I mean, we have to really rethink this. Because when you have a moron like Brendan Carr—and I say that very clearly; he’s a moron, but he’s also an unctuous toady and a thirsty person as someone who wants power—in charge of that, you get all the abuses. And of course, they do all the abuses they accused other people, the Democrats mostly, of doing, right? They do exactly—because they think, like, What’s the worst thing a Democrat could do? And really the worst thing is the things they want to do. And so yeah; I worry about that. We have to really rethink the FCC, top to frigging bottom. We’re going to have to do that.

The same thing with the Justice Department. Like, what are we going to do going forward with that? How do we reassert the powers of independence in these judiciary? What do we do when the executive loses its mind, right? And that’s been something that’s happening for a long time, the power of the exec—you know that. I mean, there’s been books and books.

Frum: I don’t think we’ve seen an executive like this ever before. And maybe that’s one of the reasons why advocates of regulation were caught unawares, because there is a long tradition of people who advocate regulation imagining a kind of benign regulator, or at least a disinterested regulator. A regulator will be someone at least non-malicious, somebody like—someone like me. You imagine the regulator in your image. And then you get, Well, wait a moment. What we’ve done is, we’ve empowered—the regulatory move has empowered Donald Trump to redistribute TikTok to his friends. Has empowered Brendan Carr.

None of the evils that you are concerned about have been ameliorated, but a whole lot of new evils that you didn’t want have happened.

Swisher: Yes, but it’s not the regulatory structure. You know, it’s sort of a perfect storm of a Congress that has abrogated its power, a Supreme Court that has abrogated—the nicest way to say it—has abrogated its power, but actually is working for the man, and then an executive willing to break laws.

One of the geniuses of Trump, I hate to say it, is like: He oversteps the line constantly. He eliminates the line. And so, what do you do when you wouldn’t imagine someone would’ve done that? And now we have to imagine, and therefore we have to put up a whole new set of lines and powers. And maybe there’ll be a much less powerful president going forward. Maybe this is the final moment of this executive theory that has been going on for, I don’t know, 100 years now—the growing power of the executive.

Frum: Let me ask you about another thing that I wonder if you’ve thought about, or rethought, since the time.

So during the Biden administration, a lot of the Biden people did want to regulate tech in some way. But they were nervous, kind of for the reasons I’ve been sort of pointing to, they were nervous about content-based [approaches]. They didn’t want to interfere with the business. And so the regulatory energy went into antitrust. It’s like, well—maybe the problem is not that the product is toxic. Maybe the problem is the company producing the toxic is too big, and we will empower Lina Khan to go after the bigness of the companies, as if that got us anywhere. And so there’s a lot of effort to curb the bigness of the platforms, without thinking about what they were doing.

How do you feel about that strategy, looking back on it?

Swisher: I don’t like any one strategy. They are too big. They’re just too big. Nobody can compete in search or in AI. It’s so expensive. It’s going to end up being two companies essentially. The redo of antitrust legislation is long overdue. I think we’re operating on a hundred-year-old law at this point. So we really have to radically rethink what big means and what the damages are. It’s not price, it’s not this, whatever. That needs a radical rehaul, and that’s something everybody knows.

The other thing is the speed at which we can meet these challenges. I mean, listen, [President] Obama, I had a long-going argument with him like: You’re not doing anything, and by the time we do something, they’re going to be too big to do anything about, because they’ll have so much power from a lobbying perspective and just money perspective that we’re not going to be able to stop them.

And then you have Citizens United on top of that, which allows these incredibly wealthy people to manipulate the system. And the numbers are just staggering—the amount how much wealthy people have put into campaigns recently. And so it’s sort of a combination of all the things at once.

So I don’t think it’s just—we have to overhaul antitrust. I really am enjoying a lot of these product-liability lawsuits that are happening in California and across the nation around these companies. It’s really a very effective way to start to go at them, is it not. We’re not going to touch free speech, but your product fucking sucks and is killing people. That’s a really good argument. So our legal system is actually meeting the moment in that regard. And there’ll be more and more. It’s like a cigarette moment. It’s happening. It really is happening. And they’re trying their best to shut it down, but they’re not always winning.

And then the third thing, which I think is the most important, is public sentiment about these people and about the things they do. And data centers—as you know, I’ve talked about, and I think I predicted this rather well—is an expression of voter discontent over all of it. And it’s just a physical thing they can point to in their own neighborhoods that they don’t like. And it is going to have a profound—it is having—a profound effect on this election. It will have a profound effect on the next presidential election, and it will have a profound effect on what we should do about the tax system. Like, that these people, they don’t pay their taxes. They don’t pay their taxes.

Frum: Isn’t the reaction to data centers kind of like the Lina Khan mistake all over again? There’s something that people really are worried about. And that’s: What does AI do to my life, to my job, to my children’s prospects? They can’t get a handle on it. And say, Well, there’s this big real-estate use outside of town. And if AI were a good thing, then this is just something we’d have to get used to, as we had to get used to gas stations. It’s part of modern life. If AI is a bad thing, then the problem is not the data center that serves the AI; the problem is the AI. And yet we have this way to go down these cul-de-sacs of bigness. Or the, I don’t like the data center, or, It’s too noisy. So we don’t talk about what’s important; we talk about what’s accessible.

Swisher: Listen, I spent a lot of time trying to find people who are trying to explain what a good data center is, too. I interviewed a Michigan mayor, and he had put together a great data-center deal. We have to—I say this to people—we have to have them. It’s a question of how we have them. And I think one of the things that tech people have tried to do is shove it down our throats without any consultation.

And I think that is what’s powerful among the voters, and I think it’s absolutely legitimate that people should have a say in how these things are going to go up. We don’t have to just, you know, bend over and take it. We don’t. We do not. And I think that’s what you’re seeing here. And it’s raising great issues of: What do we want to do about the water? And we can do very smart technological things to help save water. What do we want to do about where they’re located? What do we want to do about noise?

But instead, you had Elon Musk dropping one in Tennessee with gas belching this. You have Kevin O’Leary over in Utah acting like a giant asshole. But it was stopped, and it doesn’t mean that his data center was good. The way they approached it was in this typical—it reminds you of when the railroads went through. We’re just going to shove them right through your land. You don’t get a say. You don’t get anything. But today, I think this is a good sign of a healthy democracy pushing back, even if we actually do need some data centers in this country.

Frum: So you, you’re not only a very influential commentator on these tech issues, you’re an adept user of tech, and you’re raising a family. Do you have a personal-hygiene regime for the people you care about and for yourself to get the best out of the technology? And do you have any advice for anybody on what that hygiene regime should look like?

Swisher: Yeah, I mean, I have older kids, and they can do what they want. Like, I’m not—they’re old enough. One is 21, one’s 24. And I see young people moving very quickly away from social media. It’s really interesting. They don’t like it. I don’t mind them using the devices if they’re reading or watching substantive things. What I mind is sort of the mindless doomscrolling or the mindless silliness of it, because I try to stress upon all my kids how addictive it is, right? Because I myself am addicted. You’re addicted, I bet. It’s clearly like a cigarette. It’s got the elements of that, right? And so I like to stress with them of putting it down. My one son bought a box he puts his phone in for three hours, and he just locks it. It’s a lockbox, and it will not open no matter what he does unless he breaks it, and then he’s kind of crazy if he does that. And so I make it clear that just, it’s not controllable in a lot of ways.

The other thing I try to do is say—I’m not one of those people that says, Don’t use it. I’m like, Try it. Try it. I was just bugging my wife. She was going on about AI, and I said, You don’t use it enough to have this opinion. You need to try it to understand it. And she was like, I don’t need to try to understand it. I go, You absolutely have to try it to understand it. And so I advocate people trying it and then figuring out what they like and don’t like.

Another thing is my younger kids, they don’t use it at all. And they will not, I’m telling you, they will not have phones until they’re 16 or 17 probably. And it’ll change by then, so I don’t think I’m going to have to worry that much. But they can use it to watch movies. It’s perfectly convenient. They can use it to, you know, look up things. They do that at school. That’s perfectly fine. It’s a book. It can be used as a book. But I definitely restrict; I will. They’re not at the age where they will be using—they won’t be using social media at all. I won’t allow it.

Frum: You’re warning me that it’s addictive. But you’re also warning me that I can’t have an opinion unless I try it.

Swisher: I think you have to understand—

Frum: Do I, like Ulysses, tie myself to the mast—

Swisher: (Laughs.) Yes! Yes, close your ears.

Frum: Try it a little bit?

Swisher: Absolutely. Yes.

Frum: How do I do that?

Swisher: Well, you have to understand going in that it’s addictive, and you’ve got to figure out some strategies for yourself, like in terms of how to do it. All I’m saying is you need to understand it if you are to talk about it. You know what I mean? You can’t just be ignorant about what it’s doing.

Now, some people just, you know—like I just did an interview with someone, and she’s like, I’ve been down a glacier since 1989. I don’t know what it is. And that’s a perfectly fine way to do it if your job doesn’t depend on it, but a lot of people’s jobs depend on it.

And so I do stress that people should under—I always was, you know, I was sort of the screamer at The Washington Post, saying, Are you ignoring digital? Why are you still on a typewriter? Why aren’t you doing this, this, this, and this? And, Craigslist is going to kill classifieds. I think you need to be fully aware of the technology and understand how it can help work for you, including AI. A lot of AIs can be quite helpful in a lot of ways. And then try to minimize the stuff that is so clearly addictive and stupid and hurtful to your brain, like relationships, AI-bot relationships. Do not partake.

Frum: I was going to ask you about that. Partly because Google is so broken, I use some of the AI platforms as research tools. And I found there’s some prophylactics that I found useful. One—you can set the settings: Never call me by my name. Never tell me it’s a good question. Never praise anything I’ve done. I do not want a relationship with you. I just want to know who won the World Series in 1938. Tell me that. And I do find—

Swisher: Can I add another one?

Frum: Yeah.

Swisher: Give me the provenance of where you got this.

Frum: Yes. I tell it, these are the sources. You can’t use Wikipedia. I’ve got a list of forbidden sources—

Swisher: Or “tell me.” Tell me: Where did this come from?

Frum: Yeah, I’ve got a list of forbidden sources, and there always has to be both a site and a link, so I can check that you’ve quoted it right. Which, by the way, that’s not always true. And, you know, use government documents—even in the age of Trump—if you’re going to give me a statistic. Or, These are reputable organizations; these are not.

But what I’m mostly keen is to avoid anything where it’s a different relationship than I used to have with Google, or a box where I ask it a question and it serves me the answer.

Swisher: That’s excellent hygiene. I mean, I treat AI like a supersized Google, a better and easier Google. Because one of the things about Google is, you’d go to the page, and then you’d go down, and you’d spend a lot of time. It’s sort of like being in a library with all the books on the floor, and you’re kind of looking among them. When it’s properly sourced, it’s invaluable, like in terms of saving time on research and other things. And so I think one of the things that people have to realize—if they use it in that regard, terrific. It’s a terrific thing.

When they start to have relationships with it, which a lot of people do, or they really like it, or they answer them, there’s—like, I just did a great interview with Sherry Turkle, which everyone should read her book called Artificial Intimacy—there’s nothing there. There’s nothing there. I keep saying, I really like my… they have a name for their Claude or whatever. I was like, There’s nothing there. But they think it is because, as humans, we’re so egomaniacal, we tend to, you know, just make it human when it’s just not. And that’s a really important thing that you did.

Frum: You’re an optimist on the employment effects of AI.

Swisher: No. I don’t know. I’m—

Frum: Well, you’re not a doomer. You’re not saying that everyone’s going to be fired.

Swisher: There’s a doomer, there’s a zoomer, and then there’s—is it a boomer? There’s a doomer, a boomer, and a zoomer. I think that’s the way they’ve split among them. I suppose I’m a zoomer.

Frum: Okay, explain. You lost me.

Swisher: Okay. A boomer is, Oh, it’s going to solve everything. And that’s largely marketing from those guys. And it may. It might. It could, for sure. We didn’t know the effects of many things before they came into use. Everything’s going to be great. Nothing’s going to be wrong. It’s not dangerous at all. I just think every historical technology trend has always had an upside and a downside, period. Period. That’s the way it is if you study history. So that’s—this “up and to the right” stuff is nonsense.

There’s those who say it’s going to kill us all. It certainly could, and we should mitigate against that. You should have scenarios of consequences absolutely in place and pay a lot of attention. Nobody should ever have been surprised by the Hugging Face incident and all these incursions by rogue agents. I don’t know why anyone’s been surprised by it. It’s like—of course that’s what it was going to do. And so, the doomers can be a little bit like, There’s nothing about it that’s good. And I’m sorry, that is not the case. It’s like with nuclear energy. There’s a lot to be said about nuclear energy that’s good. There’s a lot that needs to be said that’s bad.

And so I would put myself in the “let’s go thing by thing and make a list of things we don’t want”—and I think that’s relatively easy. Which is, hey, killer drones. Maybe we should have a global decision-making on this one, with everyone deciding what the standard should be. Surveillance: Hey, let’s discuss what AI should be able to surveil and what it shouldn’t, and what parameters, and let that be monitored by government—democratic governments, at least.

There’s all manner of things we can do. And here’s a good thing. Let it help; let’s put all the data about climate change in it, and let it help us figure out solutions. Great thing for AI to do. Cancer research, that kind of stuff. You could go thing by thing.

Frum: So AI investment is what’s holding up the U.S. economy right now. Maybe also health care. We’re now on our way to a 6 percent yield for 30-year U.S. money, and it’s probably going to stay this high for a while and maybe even get higher.

How long can AI investment continue when money costs 6 percent or more?

Swisher: You know, I compare it to the internet, and remember there was a crash there, right? There was that dramatic crash in 2000. And that was nothing. It didn’t cost anything to put up these internet companies. I mean, the price that these numbers are astronomical, especially given their revenue.

Here’s all I know. There’s 12 LLMs. There’s going to be two. Inevitably, it is going to coalesce into a smaller group of people, because the business isn’t that big, you know, ultimately. And it’s not—a lot of times when these tech moguls now, because they’re searching for any way to say that it’s good after they’re the ones that told us it sucked. Which is very funny in a lot of ways when you think about it—you know, Oh, there’s going to be abundance. You don’t have to work.

Like, they’re just making that shit up, just so you know. It’s like going to Mars. Like, Oh, we’re going to live on Mars. Oh, did you know there’s too much gravity there for the human bone? Oh, no, because I talked to an astrobiologist, or We have to go underground because the radiation is going to make us dumb. So we’re going to become what? Trolls? We’re going to actually become trolls? Like, they never want to mess with the details.

Frum: Should the U.S. government nationalize Starlink and SpaceX? And if it does, how does it prevent those companies from bogging down in bureaucracy the way NASA did?

Swisher: You know, we used to have this wonderful thing in this country, the public-private partnership. Which is now gone, right? You know why you have the internet, folks? Because of the government. You know why you have a lot of the drugs you have? Because of the government. You know why you have the space program? Oh, yeah, the government.

And so I think what’s really important is, once we have something that’s a national priority—like space or, you know, AI or something—let’s have these partnerships where they don’t get to drive the car. They’re in the car with the rest of us. And what they want to do is drive the car. Because you know what, Dave? They know better than we do, because they’re smarter than us. But they’re not smarter than us. That’s the biggest lie about them. They’re not. They’re broken people just like the rest of us. And why do they get more decision-making than anybody else? That’s all. I think that’s a pretty easy thing to work through.

But I think this idea of public-private partnerships has always been one of the great things of this country. It really doesn’t happen other places. It’s either top-down like in China, and that certainly has worked to an extent. Or it’s just nothing happens, like in Europe, right? But the idea of a public-private partnership—where corporations work with governments, and some of it’s dedicated to the public good, especially when we’re paying money—seems to me to have worked for a very long time. It’s brought us vaccines. It’s brought us innovations in space. It’s brought us innovations in the internet and digital. And it could continue to do things that are great for most of society rather than a small, tiny few of them.

Frum: If you’re going to have public-private partnerships, what does governance look like? Is there a role for Congress? How does it work that we make sure you don’t get the Brendan Carrs, and telling Jimmy Kimmel not to make jokes. You don’t have Elon Musk saying, I’m blowing hot and cold on Ukraine. I like them this week. I don’t like them next week.

Swisher: I think the problem is three things that I think of. Everyone is like, Oh, it’s the internet that’s, or the tech billionaires that are killing us. I’m like, you know, it’s a really complex thing. Gerrymandering was one thing that has been killing us. We cannot have any elections that aren’t for sure, and therefore nobody has an interest in compromise, right? That’s one.

Two, social media, which has been really deleterious and politicizing people. When people have commonalities, it splits them. As its business plan is to split you from each other. And I throw in Fox News and cable with that, but I don’t think it has as much. It did have an impact; now social media has taken that over.

And the third thing I would say is, this idea that in this idolatry of these rich people, innovators, right? I think it happened previously in our history during the Gilded Age, and it caused all manner of damage for regular people, right? Whether it was children who were being industrialized into these things. And we looked at it and said, No, and we’re going to do this. We’re going to make these rules.

And I think this is the push-pull of the American system, that capitalism is a really vibrant way to move forward. But why do they get to decide?

I think we need a stronger Congress. And the only way we’re going to do that is getting rid of Citizens United, getting rid of gerrymandering. So that we have more elections where people have to compete on ideas, not on loyalty or fealty to Donald Trump or whoever the person happens to be.

Frum: Kara Swisher, thank you so much for joining me today. What a Catherine wheel of sparkling ideas, and we’ll all be listening to the show.

Swisher: You should let your AI be mean to you. That would be interesting. Be meaner. Like challenge it more.

Frum: Yes. I sometimes am mean to it.

Swisher: Oh, it doesn’t like that.

Frum: No, it doesn’t, because it has sort of biases, and then when you challenge it gets a little testy.

Swisher: Can I leave you with one thing?

Frum: Yes.

Swisher: It is us. Like, it is us. It’s trained on us. Where do you think it’s getting its information? And everything you see in our world will be reflected there. Lying, joy, happiness, sycophancy, wisdom will be all in there. And it just depends on the—again, I’ll leave you with one thing.

Many years ago at Google, they had a thing when you would go in, and it would scroll all the queries across the world that were coming in. And you’d see it. And you’d sit there, and you’d go, What are people; what are they doing? It would be like: horse, straw, France, and you’re like, What do they want? What are they looking for? Like I always sat there and I would stare at it trying to figure out what they were.

And they had a globe where all the queries were coming in, and the globe would spin, and all the different languages and everything else. It was a really great data visualization. And all the questions were light that flowed off the Earth. And when it went around, most of Africa was dark, which was really kind of this weird metaphor, right? Dark Africa. But it was because they weren’t connected, right? And they didn’t have as much connection as other parts of the world, like the United States or in China or wherever. And I looked at that, and I thought: Well, it’s not like they’re not asking questions there. You just can’t see the questions they’re asking.

And I think the ability to ask questions is the biggest ability of a democracy. To ask questions. And you don’t want your questions to be algorithmically generated. And that’s the greatest strength of humanity—is we can ask questions, we can continue to ask questions. And that’s why I don’t think it will kill us, because it can’t ask questions. It can only mimic us like monkeys.

Frum: Kara, thank you so much.

Swisher: Thank you.

[Music]

Frum: Thanks so much to Kara Swisher for joining me today on The David Frum Show.

As I said at the top of the show, this week’s book is The Rise and Fall of American Growth: The U.S. Standard of Living Since the Civil War by Robert Gordon, published by Princeton in 2016.

When the The Rise and Fall of American Growth was published a decade ago, it caused a great debate. Because Gordon’s argument, based on meticulous arrangements of large amounts of data, was that the internet revolution, social media, the computer itself had not actually done as much to improve American standards of living as the internal combustion engine, electricity, antibiotics, and the innovations of an earlier time. He argued that the American economy, or the American standard of living, grew faster from 1870 to 1970 than ever since. And all the fantastic changes in technology that we have seen didn’t add up to that much in terms of their impact on the overall state of the American economy and the well-being of the typical American.

Robert Gordon, at the time, after a lot of back-and-forth, was admitted to have probably won the argument that the American standard of living has not risen as fast since 1970 as it did before. But come the AI innovation, there is now massive challenge to Robert Gordon’s ideas. And this challenge took the form of a bet, in the world of AI—a famous bet with an economist named Erik Brynjolfsson of the Stanford Digital Economy Lab.

In 2020, Erik Brynjolfsson bet Robert Gordon. It was a nominal bet with the proceeds to go to charity, so it’s not a Polymarket Trump-era bet. This was a bet between colleagues, respectful colleagues, to say: I bet you that over the coming decade, the decade of the 2020s, American productivity growth will accelerate. And he put down a specific number. He said it will accelerate past 1.8 percent a year. Gordon took the other side of the bet, and as we stand about halfway through the course of the bet, it looks like Robert Gordon is losing the bet. American productivity has been growing faster than 1.8 percent since 2020, and especially in the past few months as AI kicks in.

So does this prove that AI is indeed a big deal, and that the Robert Gordon thesis was wrong? So two things need to be understood to understand this debate and the bet.

The first is exactly what productivity is. Productivity is a measure of output per worker, per human worker. We now have to specify per human worker. If AI were doing all the apocalyptic things that it’s supposed to do—if it were wiping out millions of jobs, if it were changing the nature of work while boosting output—that would be very obviously disturbing for a lot of people. It would change people’s lives. But from a point of view of the economic statistics, you would see that change, that revolution, in the form of a lot more output; not that much or even less employment. And so output per human worker would be going up fast.

Well, as I say, Robert Gordon lost his bet. Output per worker has been going up faster than 1.8 percent. But that 1.8 is not that great. Between 1948 and 1973, productivity output per worker rose by 3 percent a year, because of highways and antibiotics and the mechanization of agriculture. Three percent a year from ’48 to ’73. Getting around 2 percent a year now. Better than 1.8, but not great.

And all of this, I think, casts some deeper light on what AI is going to do to American work, and that’s why I returned to Robert Gordon’s book this week to refresh my mind about what he said and more importantly, why he said that. Because as disruptive as AI promises to be, most Americans work in fields where it’s difficult to accelerate work. The sectors that dominate American spending and American employment and American prices are health care, education, housing, construction, and personal care. Now, these are activities that are physical, heavily regulated, and labor intensive.

I mean, let’s say that AI is able to completely take over the job of designing new houses. And instead of architects and draftspeople and using older-fashioned computer technology, the AI designs the house from beginning to end with no human input at all. How big a deal is that? Well, if the AI doesn’t speed up the time and cost of getting a permit for the house, if it doesn’t speed up local government—simply making the designing process faster—I mean, it will disturb and disrupt the work of the people who used to do that. But it’s not going to bring down the cost of the house that much, so you won’t get a big productivity bump. And so it is with so many other areas: nursing, teaching, and so on.

When Robert Gordon warned that American growth or observed that American growth had slowed down since 1970, and he predicted that American growth would remain slow, he based that on a couple of observations about the way the American economy has changed. And these are changes that the AI revolution doesn’t seem to be undoing, not yet.

First, he pointed out that the government is much less well-financed than it used to be. There’s a much heavier debt load on the American worker—from government deficits and government debts—than there used to be. Second, we’re not raising human educational attainment as fast as we used to do. When you took the American economy from an economy where most workers had not even finished high school to one where almost everybody had finished high school and many people had finished college, that got you a big bump. In a way that the slight increase in college attainment that we’ve seen over the past years, that doesn’t change it that much. Especially when those people who do not go on to college, when there’s some evidence that their education—the quality of their educational attainment, as measured by standardized tests—that seems to have been declining a little bit since COVID and since the advent of the smartphone.

So educational attainment is not accelerating; may be stalling. The debt burden is heavy, and so much of the economy is still in the world of physical labor, physical things, regulated things. So as you observe and discuss the AI revolution, be mindful that, in the work that I do, and perhaps in the work that you do, AI may be a very big deal. But in the world of work that most people do and where most of us spend most of our money—housing, health care, education, personal care—it isn’t yet making that big a difference, and it may not make that much of a difference to the overall performance of an economy for a long time to come.

I don’t want to diminish what we are seeing. Obviously, the things these machines can do are very impressive. But where’s the benefit? How big is it? Is it really as transformational? And maybe one of the reasons that people in the AI world are so eager to tell us horror stories about the terrible things that AI might do is to compensate a little bit for some of the disappointment about the good things we want it to do but that it isn’t doing yet—and may not do for a long time to come.

That’s it for this week’s edition of The David Frum Show. See you this coming weekend. And I hope you will share the program on whatever platform; like, subscribe. And, of course, if you want to help the program and bring it to more people, the best thing to do is to subscribe to The Atlantic. That’s how you support the work of me and all of my colleagues. See you soon. Bye-bye.