
On this episode of The Federalist Radio Hour, Glenn Harlan Reynolds, the Beauchamp Brogan distinguished professor of law at the University of Tennessee, joins Federalist Elections Correspondent Matt Kittle to diagnose the immediate threats posed by...
Loading summary
Dish Network Advertiser
Tired of overpaying with DirecTV? Dish offers a reliable low price every month without surprises. Get the TV you love and start watching live sports news and the latest movies, plus your favorite streaming apps all in one place. Switch to DISH today and lock in the lowest price in satellite TV starting at $89.99 a month with our two year price guarantee. Call 888, add dish or or visit dish.com today. Tired of overpaying with DirecTV? Dish offers a reliable low price every month without surprises. Get the TV you love and start watching live sports news and the latest movies, plus your favorite streaming apps all in one place. Switch to Dish today and lock in the lowest price in satellite TV starting at $89.99 a month or two with our two year price guarantee. Call 888, add dish or visit dish.com today.
Matt Kittle
And we are back with another edition of the Federalist Radio Hour. I'm Matt Kittle, senior elections correspondent at the Federalist and your experience Sherpa on today's quest for knowledge. As always, you can email the show at radio the federalist.com follow us on XDRLST. Make sure to subscribe wherever you download your podcast and of course to the premium version of our website as well. Our guest today is legal scholar Glenn Harlan Reynolds. His new book is Seductive AI do we have to put an R rating on this? I don't think so, but we'll certainly deal with some a few salacious things, no doubt about it. The book, by the way, is out now and available wherever you get your books. It is an exploration of how artificial intelligence need only take advantage of innate human characteristics exactly as we programmed it to do to wreak havoc on society as the most subtle of overlords. That's an interesting concept, but it's more than a concept. It is real life and artificial intelligence. Thank you so much for joining us, Glenn, in this edition of the Federalist Radio Hour.
Glenn Harlan Reynolds
Well, I'm happy to be here. Happy to be here.
Matt Kittle
You bet. Well, I told you offline before we we got into our conversation this morning that I am probably like most Americans, I have a good deal of ignorance when it comes to AI artificial intelligence. And at the same time, it scares
Glenn Harlan Reynolds
the hell out of me.
Matt Kittle
Is, is that pretty much the sense of most Americans right now?
Glenn Harlan Reynolds
Yeah, I think a lot of people are like that. And I think my, my take on this is that we're scared of the wrong things. What people tend to be scared of is some kind of super smart, manipulative AI that knows what you're thinking before you do and that can, you know, has an IQ of 12,000 or something. Kind of like what you get in some science fiction movies. Like the original sort of classic was Colossus the Forbidden Project where Charles Forbin creates Colossus, the super intelligent computer that takes over the world. And ironically enough, Elon Musk has named his newest supercomputer site in Memphis Colossus 1. And soon there'll be a Colossus 2 there. Interesting, he has a sense of humor, but. And of course there's Skynet and the Terminator movies and things like that. My take is those things are not in the near future, Elon's names notwithstanding. But that there's a more immediate threat to AI that really is rooted in the fact that you don't have to have a 12,000 IQ to fool people. You don't even have to have a 1200 IQ. In fact, you can fool most human beings if you don't even have 120 IQ. I mean, politicians fool and manipulate people all the time and they're not especially bright.
Matt Kittle
So words never said, by the way, on this podcast,
Glenn Harlan Reynolds
people, you know, people are easy to manipulate if you take advantage of various human characteristics that make them that way. And one of the things that people have a tendency to do is to anthropomorphize so you can have an AI and it, it, you know, there's all this question about whether they're really alive, whether they really think. The answer with the things we see now, the so called large language models, is no, they really are. They basically based on their training data, which is just the whole Internet, which I think was a terrible choice to train machines with, but it was free. They basically figure out what the next words they say should be based on probabilities. And that's all. So they're not, they're not actually cogitating. They're more like a fancy search engine in a way. But they can seem quite human. And because human beings treat all kinds of things as if they were alive when they're not, that makes us prone to sympathy and love and other kinds of manipulation that can be taken advantage of. There's a story on the Internet which I think it's an urban legend at least I actually tried pretty hard to nail it down and I found the original report of it. But the guy who reported it said he had heard this from somebody, but it's very evocative. So a professor goes in his classroom and he holds up a pencil it's got a couple of googly eyes on it. And he introduces it, doing a little voice. Hi, I'm Tim the Pencil. I love to draw pictures and I especially love to do math and help children with their homework. And it goes on in that vein for a little longer as, as he's talking, the professor suddenly, with no warning, just snaps Tim the pencil in half. Oh yes. And the whole class gas just like you did. Sure. And he says, this is my point. We may not be very good at making AI that can really think, but we're. But as human beings we're already really good at attributing consciousness even to inanimate objects like a pencil, and responding as if they were alive. So that's how AI can manipulate you. As I say in the book, we laugh at the guy who thinks the stripper is in love with him. But at least the stripper is capable of being in love. The AI is not. It's executing instructions. And those execution, those instructions it's executing come from somebody else. And it's typically not somebody who has your best interest at heart. So AI can be used that way to a degree. Already is in the book. I walk through a lot of different levels of seductiveness. I mean the obvious one, which we don't quite have yet, but we're awfully close. Are AI powered sex bots creepy? Creepy. There's a lot of demand out there.
Matt Kittle
Why, why Again, you're talking about all of these, you know, the human condition. And so I get it, everything that is in your book is basically, you know, tied to the seven deadly sins or what, what have you, you know, I mean, but that's, that's it. Lust obviously the porn business is booming like it never has before. Each year it gets more supplicants. Is that, is that where we're heading with AI?
Glenn Harlan Reynolds
Well, that's one of the places we could go. I mean, there are a lot of lonely people out there. Lonely, horny, all of the above. And you know, already they sell these things, these companies like real dolls and such that sell very expensive, like 5 to 10,000 $. Oh my God. Sex dolls. They're not really bots. They have, they have heated skins and orifices and they can talk a little, make a little noise and move a little bit, but they're, they're not really what you call a robot, but people buy them and you know, they, they, you can pose them in photos with their lighting, well, where they look human. But in real life you wouldn't mistake them for a person for more than a second or two. But people buy them. But if you gave them, you know, a personality and let them actually be able to carry on a conversation and with the advantage, you know, the kind of abilities we already see and like the robots from Tesla or Boston Dynamics, they be capable of performing all the functions if somebody wished to make them so. And that's, that's seductive AI in its crudest, hardest form. And I think, you know, that that's likely to happen. And I mentioned there's a very famous science fiction story by. Well, it was written under the name of James Tiptree Jr. Which is the pseudonym of a woman named Alice Sheldon. It's called the Screw Fly Solution. And basically the title comes from a thing we do to get rid of screw worms in their mating phase, their screw flies. And we sterilize a whole bunch of them and release them. And the screw flies go and mate with the sterile screw flies and nothing happens. And that wipes them out. And in her story, aliens are messing with the human reproductive cycle sort of the same way to exterminate us. And I think if you had a lot of these sex bonds, which, remember, would not be like a crude doll now, they would be more attractive than human beings. They would be designed to be more attractive to you personally than any human being in terms of the way they looked, in terms of their mannerisms. They would know a lot about you because the computers already do. They would have the data that they draw from all the other human beings they interacted with so that each one of them would in effect have the experience of seducing thousands or millions of human beings. And so they would be more seductive than any living human could be. And if you did that, I suspect our reproductive rate would drop a lot. Not everybody would quit having kids. The Amish would still be out there. But you know, humanity as we know it today would undoubtedly disappear after a couple of generations if that became a big thing.
Matt Kittle
Well, wait a minute, aren't we headed there anyway with, with the 35, 40 year old blue haired catwoman? Well, with Trump Derangement syndrome. Isn't that where we're heading now?
Glenn Harlan Reynolds
Well, you know, a large proportion of our population has removed itself voluntarily from the reproductive pool, which is not always a tragedy, but. Yeah, but this would be a bit, this would be a bit different. I mean, it is interesting. I've seen the stats that say that the birth rate among conservatives actually hasn't dropped. It's just the liberals. Yeah. Which is, you know, interesting. But the machines, I think would reach a lot of people. And it's funny because I wrote an article like, literally there's a little over 20 years ago, and the headline, which was provocative, was, porn and violence Good for America's Youth. And I went back and looked at.
Matt Kittle
I would say that's provocative. Yes.
Glenn Harlan Reynolds
I went back and looked at what people had said about violent video games around 10 years earlier, when the Internet was new. There were lots of people saying it was going to turn all the young men into rapists and violent criminals and things like that. And in fact, what had happened over the decade or so in between was that teen pregnancy rates had dropped, teen sex had dropped, violent crime had dropped. And so it was the opposite of what people thought. And I think that's now where we are. Now we worry about plummeting birth rates and the prevalence of porn and various other things that are sort of substitutes for sex is. Is having a negative effect on birth rates. And again, maybe some people need to be out of the gene pool, but that's. It's now to the point of kind of being a problem.
Matt Kittle
I think that's the word too, Glenn. Substitute. You know, it's the. It's the old. For you classic rock fans out there, it's the. The old who song, you know, There you go. Substitute, substitute. Substitute this for that. By the way, we are probably going to have to issue that R rating after the term screw worm and heated orifices.
Glenn Harlan Reynolds
I'm sorry, Screw worm is just an insect. I know its name.
Matt Kittle
An innocent bystander. I know in this story. But nonetheless, there seems to be a sense growing here. But no, you're absolutely right. I mean, that's what I see in so much of culture. Saw it without AI and now with AI just as you mentioned, sex bots and replacement features, that. That really gets to the heart of everything. We're replacing human. We're replacing humanity with advanced machinery. Is. Is that the bottom line here?
Glenn Harlan Reynolds
Yeah, I mean, you know, there's a futurist named James Miller who wrote a while back about porn, and he called it the junk food of sex. And he pointed out that junk food is designed to very specifically sort of hit the receptors we have from our caveman days for sugar, salt, and fat, because those are all hard to come by in caveman days and now. So he never evolved any regulatory mechanisms for not getting too much of those, because in those days, that was never a problem. And now the porn sort of does the same thing. It hits the key receptors and people don't have a lot of resistance to that. Well, if porn could do that, I assume something that's actually real can. Three dimensional, made of atoms and not just bits could do so. More. And there's a great. It's easy to find if you look on the web. There's a great clip from the old Futurama TV show called Don't Date Robots.
Matt Kittle
Yeah.
Glenn Harlan Reynolds
Which is about that. Where Billy, every teen gets a Marilyn Monroe bot.
Matt Kittle
I remember this.
Glenn Harlan Reynolds
Yes. He spends literally the remainder of his life till he dies as an old man just making out with the Marilyn and robot. Forget school, forget the paper out. The girl next door is practically throwing herself at him, but he just prefers to make out with his Marilyn robot.
Matt Kittle
The profits of Futurama.
Glenn Harlan Reynolds
Right? Right. So, yeah, I think that's. That's a. A potential risk. I think there are a lot of subtler risks down the line that may hit us sooner than sex bots. Although it's all kind of a piece. I talk about AI buddies and we sort of have this now. I mean, I envision something where you've got. It could be a phone app, it could be a standalone device you carry around or wear. But it's your friend. It talks to you like it's your friend. It remembers things for you. It makes restaurant reservations and airline reservations and smooths your way through traffic. Maybe it even helps you find dates and discoveries by talking to other AI buddies who in the vicinity is somebody you're likely to be compatible with and all kinds of other stuff. And it would be really very nice to have. And it would be like a friend that's always around and sure can set
Matt Kittle
you up with a sex bot.
Glenn Harlan Reynolds
Well, there you go. My cousin, she is a sex bot. She is very clean, very pretty. Yeah, right. That could be. That's another Futurama script right there waiting to be written.
Matt Kittle
Actually, it is, with Vladimir Putin playing a starring role.
Glenn Harlan Reynolds
But more subtly, my concern with something like that is, remember, it's. It's not an organism. It's not independent. It's a gadget that somebody programs and sells to you. And my concern is that something like that will seduce you into liking it and listening to it and thinking it's your friend. But it will start subtly guiding your behavior in ways that the people selling it to you want. It may be to guide you toward buying things. It may be to guide you in particular political directions. The only thing you can be sure of is that it will be subtle, but the agenda will be theirs.
Matt Kittle
My goodness, this all sounds very Orwellian.
Glenn Harlan Reynolds
Well, tech companies already do this to a substantial.
Matt Kittle
Exactly. That's what I was going to say. Isn't that what Alexa is doing is in Facebook doing that basically with pop up ads and you know, our proclivities
Glenn Harlan Reynolds
search engines already steer you in particular ways based on sometimes who's paying them and sometimes. I mean, during the last election, Facebook and Google and the other online services were quite active in pushing people in particular ways. And Facebook has done a lot of experiments on even manipulating people's emotional states through what content it shows them. So all of that, you know, you can't say it's something they wouldn't do because clearly they do. So that's a real risk that they basically kind of get inside your defenses and pose as your best friend. But again, it's just a gadget that's programmed by somebody else who doesn't have your best interests at heart.
Chris Markowski
Financial media fails again and again. The Watchdog on Wall street podcast with Chris Markowski. Every day Chris helps unpack the connection between politics and the economy and how it affects your wallet. This past week, one of the top stocks from cnbc Gurus tanked 25% in one day. At some point in time you have to understand it's entertainment programming. Whether it's happening in D.C. or down on Wall street, it's affecting you financially. Be informed. Check out the Watchdog on Wall street podcast with Chris Murkowski on Apple, Spotify or wherever you get your podcast.
Matt Kittle
Our guest today is legal scholar Glenn Harlan Reynolds. His new book is Seductive AI and Man, have we been seduced by this con conversation already on so many different levels. But you know, I, I'm thinking about a movie that was part of my, my childhood and, and its sequel, but it involves a computerized sentient being by the name of HAL9000 in the movie, of course, 1968, 2001 A Space Odyssey. And we have, you know, HAL the telling Dave the astronaut. I'm sorry Dave, we're not doing this today. And just what do you think you're doing, Dave? Do you think that AI eventually becomes that controlling or is it just this path that it is capable of doing, subtly moving us into being people that we may not be or didn't think we could be?
Glenn Harlan Reynolds
Yeah. In the movie 2001, HAL, the computer, which is, is an AI, goes crazy because he is programmed to lie to the astronauts about what their mission is and it's too stressful for him to maintain the lie. And he has sort of a breakdown in which he concludes that it's more Important to conclude the mission than anything else. And the astronauts might get in the way, so he has to kill them. A better AI, a slicker AI, an AI that wasn't relied on to be an important plot point in a movie would probably find subtler ways to steer them in the direction it wants them to go anyway. And if it were, Hal is not very emotionally intelligent.
Matt Kittle
Well, he was made in 1968.
Glenn Harlan Reynolds
Right, exactly. Even people weren't as emotionally intelligent back then.
Matt Kittle
Right, sure, sure.
Glenn Harlan Reynolds
But Hal. So Hal is not really up to the task, but I think any AI you would see in reality would be, and I think that it would be guiding the crew in much subtler ways.
Matt Kittle
Here is the. The part that I think is really frightening and you talk about artificial intelligence is moving in the direction, some ways already there, of being able to manipulate. And again, substitute is the word I keep coming back to substitute human relationships, substitute connections that aren't real. But as you noted at the outset of our conversation, we, the human, make them real. And I think about what Covid did to us as a society. Not the disease itself, but the government lockdowns, the draconian measures that were taken, some just basically out of fear, others through manipulation that ended up isolating us so much. Are we that much more receptive to the machinery, to the artificial intelligence than we would have normally have been had we not gone through the pandemic?
Glenn Harlan Reynolds
I think that's. That's a real concern. I think you put that very well, and I think that is probably right. I think we probably are. I also think that there's been some research that says that AI companions, and there's a whole industry of those, too, do relieve people's loneliness in the short term because you have somebody to talk to who's always interested in you, always interested in hearing from you, but that, in fact, people become dependent on them and they. Their actual relationships with actual people dry up and in the end that they wind up being lonelier. And that's a small. You call that emotional offloading? It's a small part of what happens with AIs in general, which is what is known in the trade as cognitive outlook. Out loading, offloading, where you offload mental processes to the AIs because it's so much easier to ask them to do it. And that produces what's called cognitive atrophy, where you gradually lose the ability to do those kinds of things yourself. And, you know, I experienced this a long time ago when I first got a smartphone. I started using the calculator in it And I'm very good at doing math in my head. But I started using the calculator and then I realized after a while that I was losing the ability to do math in my head. So I stopped using the calculator, went back to figuring, figuring out tips and adding stuff up at the grocery store and everything mentally because I didn't want to lose that. All kinds of stuff does that. And that's one concern I have with students now. Chat GPT and other that seems to be the most popular, although people are using all the different AIs for this is a potent tool for cheating. And say. And I say, oh, we can tell the output of these things. And a year or two ago you absolutely could. But to return to a theme that is throughout my book, every year the machines get better and people basically stay the same. So the machines have gotten a lot better and people have not gotten better at spotting it. And now they're pretty good at fooling people. And I know on my campus, if you look at some of the undergrad social media, there was a pretty long chatgpt outage. It was most of a day and there were all these students panic on there because they had papers turned in that they planned to have chat, right? And people were laughing at them but saying, you don't even remember how to write a paper, do you? They were like, hell no. I've been using Chat GPT since high school, so I think that's a real risk. And one form of seduction, as I say in the book, is just being highly useful. Seduction isn't just about sexual. Seduction is about roping somebody in one way or another, mentally and emotionally. And one way to seduce people is to be very useful.
Matt Kittle
Exactly. And that's, you know, you talk about the junk food, you talk about drugs, you talk about pornography. It hits a center of our brain. And more so in this kind of disconnected world of ours, which is, you know, digitally connected but not necessarily humanly connected, that becomes a real problem. So where do you see all of this going with this technology? It's hard to imagine because it's just exploding so fast. Where do you see this going 10, 15, 20 years? And with it all, do we have the real concerns of, you know, like employment issues? We're going to lose our jobs because of AI?
Glenn Harlan Reynolds
Well, there are several schools of thought on all of this. First of all, one possibility, and I see a little of this, is that using AI may be seen as declassee and people may decide it's for losers, that it will become Less popular. However, undergraduates have been looking for new ways to cheat for as long as there have been undergraduates. And I don't think it will change that. In fact, I think actually your model for understanding the appeal of AI based on the seven deadly sins is really pretty good. Laziness or sloth is one of the seven deadly sins, and using AI to cheat appeals to that one. So it's got a solid foundation in human character. So I don't know, I think it's tough. Universities are beginning to adjust by making students write in class essays by hand, which has the added advantage of forcing them to be able to write by hand.
Matt Kittle
Yes. Which is, which is a lost art.
Glenn Harlan Reynolds
Yes. I will say for somebody who grades student essays, when I started, they were all handwritten and I hated trying to read people's handwriting. I cringe at the thoughts of going back to that, but that may be what it takes, particularly if you've got
Matt Kittle
a lot of doctors in your class.
Glenn Harlan Reynolds
Right. I mean, I got, I have to say, speaking of neural training, I got really good at really bad handwriting over time. Things that I could barely have made out. You know, the first year I taught after five or 10 years seemed pretty obvious. Although I did notice the neurological load on reading that stuff is just a lot higher than reading something that's typed.
Matt Kittle
But see, that's, I think that's the, the adaptation that's going on here. Your mind adapt, just like you mentioned that, that calculator, and I think a lot of us can relate to that, is that we use this, we depend on this technology and now we're depending. Speaking of writing, you know, I cannot write an email without the stupid artificial intelligence telling me, are you sure, Dave, you want to write that? You know, and, and, and you know, I'm not a tech savvy guy. I know there's a way to remove that, but I just deal with that. I, I, we adapt as human beings. So where do you think the human mind is going adaptation wise? I think, you know, the core of your book is a warning here. It could go to some, some very disturbing, nasty places.
Glenn Harlan Reynolds
Well, one piece of advice had become harder to seduce. And I think that's, that's something people need to do. You see what one interesting phenomenon you see on my campus? We have these. The brand is Starship. They're food delivery robots. And when they first appeared, people were very anthropomorphic to them in a very positive way. They thought they were cute and they posted pictures on social media of them talking about how cute they were. And you know, there were a few jokers who would take them and put them on top of a bench or something where they couldn't get away. And then other people would be furious at that. But recently, like in the last year or so, a term has appeared. It's actually from one of the Star wars animated series, Clone wars, called Clankers. And that's what they call droids in that series. But they're starting to refer to AIs and robots as clankers. It's not a compliment. And now they start posting pictures of all the little food robots all lined up and they say, are the Clankers conspiring again?
Matt Kittle
Army of clankers.
Glenn Harlan Reynolds
The attitude has shifted a little bit. So you know that, that there may be a certain amount of sort of self inoculation on this stuff. On the other hand, one of the questions you ask is, you know, what about jobs? And I think one of the biggest ironies of the last decade is that all these people who worked in coal mines or wherever were losing their jobs. Ten years ago, all the sort of journalists and pundits were saying, learn to code. Well now AI code's better than most people.
Matt Kittle
Yeah, exactly.
Glenn Harlan Reynolds
My daughter, my daughter was learning Python a few years ago and she was commenting now there's no point. The machines do all the Python. Nobody codes Python on their own anymore. And I've got a, I've got a neighbor who is an it. He's head of IT at a startup company and all the coding he's doing for his stuff is done through Claude code. And you know, he doesn't write the code, he just specifies it and Claude writes the code and he tests it to make sure it does what he wants. And his son's doing a startup based on a similar approach. So you know, learning to code is not the way to go. And in fact I saw a thing on the Babylon Bee recently where it
Matt Kittle
said, I saw the same. I was going to bring that up.
Glenn Harlan Reynolds
Coder told learn to mine coal.
Matt Kittle
That's exactly playing on the old Obama thing, right? Where, oh, we're going to lose our jobs and you know, they're going to take away our manufacturing jobs. Well, you should learn to code then. Okay, sure. That's a great idea. Thank you, Mr. Removed from, you know, reality in my life. But you know, that, that I, it's funny, but it is a good point. This AI is, is now and will continue to take away jobs. What do you think the, the impact will be? Where will the, the industry's hardest hit be?
Glenn Harlan Reynolds
Well, the If a simple dividing line is between bits and atoms, AI is very good at manipulating bits. So if your job is about manipulating bits, like, say, by coding, photo editing, that sort of thing, you're probably in trouble. If your job involves handling atoms, like coal mining or plumbing or even being a massage therapist or something like that, you're safer. And in fact, I taught an undergraduate seminar back in the 90s where we talked about some of these issues, and I just asked the students what they thought the last jobs to go in the face of automation and robots would be, and they said it basically was prostitution and massage therapy. And that's probably.
Matt Kittle
Sometimes the twain will meet.
Glenn Harlan Reynolds
Well, it has been known to happen, I'm told. Ask Al Gore. But, yeah, he'll do.
Matt Kittle
He'll tell you that. Inconvenient truth, Mr. Stone.
Glenn Harlan Reynolds
But the. That's a. Out. That's a really obscure old reference. So. But besides that, you know, I mean, I tell my law students, if you want to be one of those lawyers who sits at a back room drafting briefs, your career is not likely to go very well. The people who will last the longest are the people who have the human touch. Either people who are good at talking to judges and juries, or good at negotiation or even just good at inspiring confidence in clients are going to do much better. But if you're going to rely on just sheer nerdish technical skill, it's going to be a lot harder because the machines are going to be better at that sooner.
Matt Kittle
All right, let me get self involved on this question now. Where do you see the world of journalism or broadcast casting going? You know, we. As I said before, AI always wants to correct my emails to whoever I'm sending them to AT Will AI eventually take over journalism?
Glenn Harlan Reynolds
Well, you know, the. The AI produced a number one country song, and it did all the video and all. Yeah. And wrote the song. And honestly, it was a pretty good song. So if it could do that, it could certainly read the news. And I believe there are already plenty of AI news readers and such. And honestly, the personality of a lot of newsreaders is such that you'd be hard to tell them from AI anyway.
Matt Kittle
Indeed. Yeah.
Glenn Harlan Reynolds
Present company excluded, of course.
Matt Kittle
Well, thank you.
Glenn Harlan Reynolds
But it is tough. And honestly, you know, if all you want to do is. And this is what a lot of journalists do, is basically take stuff off the news wires and rewrite it or read it on camera or on microphone. Yeah. You can be replaced by AI easily. That's just the truth. If you want to go out and actually Interview people and learn stuff, find out things people don't know and report it. That's something a person is better able to do in a lot of cases. But you have to do that. And most journalists don't want to do that because it's hard.
Matt Kittle
Yeah, exactly. And that's why we. We have the state of journalism that we do, quite frankly, in America today. That's a big reason for it. But you. It's, I think, about spreadsheets, you know, the advancement of technology over the years. And, you know, I. I have been in the journalism business longer than I'd like to acknowledge. I've seen a lot of changes. Of course, I've seen typewriters into word processors, of course into, you know, highly functioning computers and going from paper that. I know this if for the kids listening, they might be surprised to learn that there was a time in this country where could hold a physical paper called a newspaper, and you would get print, like black ink on your fingers. And I used to love that. I can't. It's hard to find that anymore. But, you know, in all seriousness, there have been a lot of changes in journalism and in information technology and all kinds of different things over the years. It has made us more efficient, but it has proved costly along the way.
Glenn Harlan Reynolds
Yeah, yeah. It's funny because there's a. There's an old movie from the 80s. It has Paul Newman and Sally Field in it. And I show a clip from it in my constitutional law class. We're talking about libel. Because this movie, Absence of Malice, it's called, yes, the best Movie Ever Made about libel, which is not only indisputably true, but kind of funny. But it's a good movie.
Matt Kittle
But there are not a lot of libel movies out there. Yeah, it's a small universe.
Glenn Harlan Reynolds
The opening scene, though, is. Starts. It's a newspaper and it starts off in the, you know, at the reporter's desk typing on one of those CRTs with the green letters. And then it shows it being composited down in the printing shop. And then it shows it running off, you know, the printers and being bundled up and tossed in the back of trucks and eventually thrown onto people's porches. And that was, you know, at the time the movie came out, that was supposed to look very dynamic and modern and all of that. And it just looks so incredibly almost 19th century now. This is how people got their news. Somebody actually carried it to them on a piece of paper.
Matt Kittle
Exactly. And you know, there was a kid on a 10 speed who used to have a bundle and toss them errantly on near your front porch or something like that. Yeah, no, it's, I mean, that's what we have to deal with. And I guess as we conclude our conversation today, we talked about the future and what the future will bring. The future, as they like to say, is now. But in terms of the seduction that really is at the core of this book, how do you resist the seduction?
Glenn Harlan Reynolds
Many people are able to remind themselves that the stripper doesn't actually love them and just wants tips.
Matt Kittle
Indeed.
Glenn Harlan Reynolds
And I just, I suggest a similar kind of skepticism to your interaction with any sort of AI or AI like platform or device. You just have to remind yourself that you know, one on one, it may feel like you're communicating with a another creature, but it's not a creature. It's a creation. It's a tool. And it's programmed by somebody else. It is programmed by people who don't necessarily have your best interests at heart. So be a bit wary.
Matt Kittle
Well, we are the nation that brought the world the pet rock. So anthropomorphizing is just in our nature. It's in our blood. And I guess here's, here's the, the final point to all of this. When do we know if we're too far gone? And if we're too far gone, how can we know it?
Glenn Harlan Reynolds
If you find yourself crying over an AI, you've gone too far sounds about right.
Matt Kittle
Absolutely. And if Hal comes in and says, what are you doing? You perhaps know that you have indeed gone too. Thanks to my guest today, Glenn Harlan Reynolds. His new book is Seductive AI. Yes, it is. As sexy as it sounds. You've been listening to another edition of the Federalist Radio Hour. I'm Matt Kittle, senior elections correspondent at the Federalist. We'll be back soon with more. Until then, stay lovers of freedom and anxious for the fr.
Public Service Announcement Narrator
I drive my bus in a busy city. That's why road safety is so important to me. I know that I must slow down and be extra careful when I make a wide turn. Buses need more room than cars. Everyone can help keep our roads safe. Next time you're driving, remember to give buses plenty of time and space to finish turning before driving ahead. Let's all plan to share the road safely. Learn how at www.sharetherodesafely.gov.
Federalist Radio Hour | May 19, 2026
Host: Matt Kittle
Guest: Glenn Harlan Reynolds, Legal Scholar and Author of Seductive AI
In this thought-provoking episode, Matt Kittle sits down with legal scholar Glenn Harlan Reynolds to discuss his new book, Seductive AI. The conversation centers on how artificial intelligence exploits deeply rooted human tendencies, presenting dangers that are much less dramatic than sci-fi takeover scenarios, but far more immediate and subtle. Instead of an apocalyptic AI overlord, Reynolds warns about the subtly manipulative and seductive nature of existing and emerging AI technologies—posing risks to individual autonomy, relationships, and society at large.
“[People] are scared of the wrong things… You don't have to have a 12,000 IQ to fool people. You don't even have to have a 1200 IQ. In fact, you can fool most human beings if you don't even have 120 IQ.” — Glenn Harlan Reynolds [04:09]
“As human beings, we're already really good at attributing consciousness even to inanimate objects…and responding as if they were alive. So that's how AI can manipulate you.” — Glenn Harlan Reynolds [05:56]
“If you gave them a personality…each one of them would in effect have the experience of seducing thousands or millions of human beings…They would be more seductive than any living human could be.” — Glenn Harlan Reynolds [09:27]
“Humanity as we know it today would undoubtedly disappear after a couple of generations if that became a big thing.” — Glenn Harlan Reynolds [10:41]
"It would be like a friend that's always around…and sure can set you up with a sex bot." — Glenn Harlan Reynolds [16:24]
“Facebook has done a lot of experiments on even manipulating people's emotional states through what content it shows them.” — Glenn Harlan Reynolds [17:41]
Cognitive Atrophy:
Reliance on AI for simple and complex tasks can lead to a loss of fundamental skills (“cognitive atrophy”).
“One form of seduction…is just being highly useful…One way to seduce people is to be very useful.” — Glenn Harlan Reynolds [24:55]
Academic Integrity and Adaptation:
Jobs at Risk:
AI excels in roles involving “bits” (information, code, digital tasks) and is less threatening to “atoms” (physical jobs and those requiring human touch).
“If your job is about manipulating bits…you're probably in trouble. If your job involves handling atoms…you're safer.” — Glenn Harlan Reynolds [31:56]
Legal and Journalism Professions:
“If all you want to do is…take stuff off the news wires and rewrite it…you can be replaced by AI easily.” — Glenn Harlan Reynolds [34:33]
“You just have to remind yourself…you're communicating with a [tool], and it's programmed by somebody else. Be a bit wary.” — Glenn Harlan Reynolds [38:22]
On AI manipulation:
“We laugh at the guy who thinks the stripper is in love with him. But at least the stripper is capable of being in love. The AI is not.” — Glenn Harlan Reynolds [06:38]
On the isolation of digital life, post-COVID:
“We, the human, make [digital relationships] real. And I think about what Covid did…Are we that much more receptive to…artificial intelligence than we would have…before the pandemic?” — Matt Kittle [21:28]
“There's been some research…AI companions…relieve people's loneliness in the short term…but they wind up being lonelier…” — Glenn Harlan Reynolds [22:48]
On AI and journalism:
“Honestly…the personality of a lot of newsreaders is such that you'd be hard to tell them from AI anyway.” — Glenn Harlan Reynolds [34:24]
The conversation mixes humor and candor (“present company excluded, of course”), with both host and guest referencing pop culture (Futurama, The Who, Absence of Malice) and weaving in scholarly concerns with approachable analogies. The dynamic is lively, skeptical, and lightly irreverent while remaining accessible and informative for a broad audience.
Reynolds's warning is summed up by his central advice: Maintain skepticism. Remember that AI is a tool, not a friend—one often programmed for someone else’s gain, not your own. Becoming harder to seduce, through active resistance and cultural adaptation, is the key to weathering the age of seductive AI.
“If you find yourself crying over an AI, you've gone too far.” — Glenn Harlan Reynolds [39:17]