I'm Kevin Roose, a tech columnist at the New York Times.
I'm Casey Newton from Platformer, and this is Hard Fork. This week, prediction markets are out of control. Is Congress about to rein them in? Then Joanna Stern returns to the show to discuss her new book on turning her life over to a chatbot. And finally, Hard Fork's own Rachel Cohn returns to the show to talk about her first month at attention school. She has our full attention. She does.
Well, Kevin, a few weeks ago you predicted we would soon do another segment on prediction markets, and I'm happy to tell you that prediction has now come true.
Oh, thank God. My bet is going to pay out on Kalshi.
It is, because as I was looking at the news of the week, it seemed like everywhere I opened up a browser tab, Kevin, a prediction market had been in the news, often not for a great reason.
Yeah. I mean, this has been one of the tech stories of the year, just the absolute meteoric rise of prediction markets in the popular imagination. I've been walking around New York for the past day and just like ads for these prediction markets are everywhere you look. It has like taken over culture in a way that I'm not sure I would have predicted.
Yes. And one way that prediction markets keep entering the news, Kevin, is it seems like every other day I am reading a story about a massive insider trading scandal that has unfolded on one of the platforms.
Yes. So, you may have seen about two weeks ago, we learned about an army sergeant who was allegedly involved in the capture of Venezuelan President Nicolás Maduro who made more than $400,000 placing bets on markets related to Maduro being out of power by the end of January.
Yeah, not great. And he is not a total outlier. A group called the Anti-Corruption Data Collective analyzed more than 400,000 prediction markets settled on Polymarket over the last five years, and they found that longshot bets related to military or defense had an average win rate of about 52%. Now keep in mind the average win rate on this platform is 14%. So if you go and you see a big bet on one of these sites about the military, somebody might be betting on information that they really should not be.
Yeah. I mean, this just seems like something that is obviously more widespread than we know about. Like, if you have material non-public information about a military operation, like what are you going to do? Sit there and collect your freaking paycheck like a chump? Are you going to go online and make some dough betting on the outcome?
You know, I remember, you know, the app Strava, which kind of like logs your runs and your bike rides. They got in trouble once because they were publishing these heat maps which inadvertently revealed the locations of some US military bases. So they had to shut that down. Fast forward a few years later and now the sergeants are just placing bets on operations that they're actively involved in. Um, you know, another great insider trading scandal, I wonder if you saw, Kevin, took place in France where a police complaint was filed by the weather forecasting service alleging that its equipment for measuring the temperature at Paris's Charles de Gaulle airport was interfered with, which coincided with a surge in suspiciously well-timed bets on Polymarket. I loved this one because my understanding, and correct me if I'm wrong, is that there's this prediction market for like what is the temperature in Paris, and the way that they gauge this is with this series of thermometers that are placed in various parts of Paris, and that this insider trader allegedly basically took a hair dryer or some other heating device and held it next to one of these sensors.
Okay, so let me... can you just tell me what happened here?
Yes. So this was also my understanding of what had happened until I looked into it. And it turned out that while there is an allegation that these sensors were tampered with, the photo that was circulated of someone holding a hair dryer up to the sensor had been generated with AI and was circulating in one of the Discords for one of the prediction markets. So it's not just a story about prediction markets. It's also a story about slop and disinformation. I fell for that one.
Welcome.
So, how did they actually tamper with the temperature sensor?
That part is still unknown, but what we do know is that on April 15th, the recorded temperature jumped at Charles de Gaulle from 18 Celsius to 22 C. So, you know, this just feels like an incredible crime of opportunity to me. You know, like if you could just walk up to a thermometer with a hair dryer and make yourself $14,000, you might do it, knowing you.
But this is a problem, Kevin, because not only are people essentially like defrauding the other people who are participating in these markets, but I just think it's really bad for the markets themselves because they have pitched themselves as these miraculous systems for discovering the true price of things and harnessing the collective wisdom of the crowd to help us understand current events. And everywhere we look around, we see that the people who are making money appear to be manipulating the markets in these very devious ways.
Totally. And I think that is ultimately bad for the markets themselves. Market integrity is obviously very important. If people start to feel like they're competing on these markets with people who have access to like, you know, insider information, um, that's going to dissuade them from doing it. I mean, I was thinking about this after the Bad Bunny halftime show at the Super Bowl, um, where there were lots of prediction markets, uh, on what songs Bad Bunny would perform and various other celebrities will appear.
Celebrities will appear and there were active prediction markets. And it turned out that like probably some of the people betting on those markets were like part of the halftime show or had watched the rehearsals or something. And it just feels like after enough of these incidents, like you kind of have to be a sucker to participate in these markets without insider information. And like what happens if that goes, if just the normal people who just want to go online and gamble a little bit of money on something go away because they think it's rigged.
Absolutely. And by the way, I have to say after that halftime show, I got so into Bad Bunny. Me too. I don't care that I'm the last person to figure this out. Okay, Tití Me Preguntó, incredible song. It's a bop. It's a bop. Okay. But to the exact point that you just made, most people who bet on prediction markets lose, right? Um, according to the Wall Street Journal, which did some great reporting on this over the weekend, on Polymarket, more than 70% of users lose money on the platform. And at Kalshi, there are 2.9 unprofitable users for each profitable one based on data from the past month. So, I think these are just important things to keep in mind if you are walking around New York City and you happen to see a lot of ads for these platforms and you think, 'Hey, I'm going to go turn a quick buck.' Like, at the very least, know that the odds are against you.
Yeah. I mean, it speaks to the reason why we have insider trading laws for stock markets. It's not just because when you insider trade, you are like depriving someone else of money. It just makes the whole market less fair and it destroys the trust in the market that makes it possible for it to be liquid and transparent. So I think these insider trading scandals just show like right now we are sort of at a pre-regulatory wild west moment for these prediction markets. I imagine that will change at some point because they don't seem like they're going away and we just kind of need someone to step in and like say okay we're going to establish some rules so that we can like protect the integrity of these markets.
Yes. Well, and there have been increasing efforts to try to regulate these platforms, which we should talk about. Interestingly, a number of states have now tried to intervene saying, 'Hey, we want to ban this stuff in our state. We don't want this.' So, the Commodity Futures Trading Commission or CFTC has actually sued these states and said, 'No, no, no. This is our exclusive domain. We are the ones who get to regulate this.' And also, by the way, we don't really want to regulate this. So, tough, tough beans for you. So that's sort of been frustrating if you're on the side of somebody ought to do something about this.
I mean, I think there's a couple systemic issues here. One is that the CFTC is just quite small. The CFTC relative to the SEC, which regulates the stock market, is just like a tiny fraction of the enforcement team. It was not really meant to regulate prediction markets. It kind of ended up there sort of via this historical accident where like Kalshi was doing these things that were technically considered futures contracts which brought them under the jurisdiction of the CFTC. I think there's a real argument to be made that like as this stuff gets more widespread, it should move toward something like the SEC, which just has a lot more resources to investigate insider trading.
I wouldn't be surprised if the prediction markets weren't lobbying to continue to be regulated by the CFTC because we saw the crypto people do the exact same thing. They said, 'We don't want to be regulated by the SEC. They're really good at their jobs. Let the CFTC do it.'
So, here is maybe the good news if you're hoping that there will, you know, get some adults in the room here. The Senate unanimously passed a rule barring senators from betting on prediction markets. Finally answering the question once and for all, Kevin, will the Senate ever do the bare minimum? They did.
Um Kevin, please don't get way ahead of yourself. We have to see if we destroy capitalism by preventing the senators from betting on prediction markets.
Can Supreme Court justices bet on the outcome of Supreme Court cases?
You know what? I bet when they do, we're going to hear about it in ProPublica. They seem very good at that sort of thing. So there's a little bit more action here in the United States. Two US senators, including Kirsten Gillibrand and Dave McCormick, have now introduced a bill that would ban members of the legislative and executive branches from trading on prediction markets. So you know that would presumably prevent the president from betting on prediction markets and that's something that he's been considering. And we're also seeing some action in other countries. Brazil has now blocked 27 sites including Kalshi and Polymarket for offering what they're just calling illegal gambling. France and Hungary have banned them as well. So Kevin, this just sort of seems like once again a case of the rest of the world being like this thing that seems bad, we're going to put a halt to it. While America says, 'No, no, my friends, for there is money to be made. Go forth and make it.'
It's really this topic is so interesting to me because do you remember like when I went to that prediction markets conference and like I, you know, I'm not a guy who likes to do sort of like remember when I saw Green Day at the corner bar and they were playing for 16 people and you know look how cool but like I do feel like I saw the equivalent of Green Day playing the corner bar like the people who were interested in prediction markets several years ago were like these absolute like nerds in the Bay Area who were sort of involved in the kind of play money prediction markets. They were not like businesses that had like billions of dollars. It was like this very niche academic interest. And I remember going to that and feeling like I'm not sure whether this should be legal or not, but if it ever is, like I imagine this is just going to become like a total casino. And I remember arguing with someone there about insider trading. And this person who was one of the people who were sort of originators of this movement were like saying that insider trading is good in a prediction market. You want insiders to be trading on these markets because that produces better information. And the point of prediction markets is to produce better information. And so if you have members of Bad Bunny's, you know, entourage betting on the Super Bowl or you have people betting on military operations that they're actively involved in, that is actually a net good because then we're more likely as a society to know that something is going down in Venezuela or something is happening at the Super Bowl. And I just remember feeling like that is a beautiful theoretical construct that has zero chance of surviving contact with the real world. And as it turns out, it didn't survive contact with the real world.
No, because it turns out what you are incentivizing everyone in the world to do is just to betray those closest to them. Yes. Like betray your friends, your family, your co-workers, your country. Just do it all for a quick buck.
So, I think we should sort of take this to what do we do about it, Kevin? And I'm curious what, if anything, you think we should do? I mean, I just think this is one where we just need a new way of regulating these. Like, right now, these companies are self-regulating. You know, Kalshi has said we don't allow insider trading. We don't allow death markets, which is basically betting on the death or assassination of a public figure because that could incentivize someone to like go out and kill the person, for example, to claim the bounty. So, they are instituting these rules unilaterally for themselves. But that seems like step one.
Yeah, I think there's kind of two big categories of harms here that just have to be addressed differently. There's a set of harms related to gambling, right? Like some people become addicted to gambling and I think these prediction markets are set up such that people could develop that kind of problem. And so I think this industry needs to be required to do the same sorts of things that casinos do, which is you have to let people exclude themselves from the market. If they say, 'Hey, I can't trust myself with your particular prediction market.' I think they need to do mandatory age verification, right? I don't want to read a story in a year about like the high schools where Kalshi is the hottest thing and you know there's a bunch of 16-year-olds in debt because they couldn't stop betting on who is going to be in the Super Bowl. And then I think we probably need to have some limits around advertising. I don't think blanketing the world in advertisements for gambling is like going to lead us to a good place. But then you also just have the market problems which is what you're talking about which is that clearly insider trading is just an inherent feature of these platforms. And so we do need a big bad regulator that is just actively surveilling these platforms and is trying to get the bad actors off the platform. And if I were a Kalshi or a Polymarket, I would welcome that because then my prediction market might actually be worth something, you know, because it wouldn't just all be people, you know, holding up hair dryers to the temperature sensors at Charles de Gaulle airport, which didn't actually happen.
Yeah. And I would like to see prediction markets become something closer to the vision that I heard back at that prediction markets conference years ago which is like a way of sort of incentivizing the production of good knowledge. To me, one of the things that the proponents of prediction markets were saying is like right now we have polling for like public sentiment or elections and people are not incentivized to like go out and do their own polls because they think they can do a better job than Gallup or you know Ipsos or whoever the sort of polling organization is. But if you have prediction markets where people are like incentivized to go out there like do their own polling, do their own research because it might help them make money that's going to create like a more flourishing system. And like I would just like to see that kind of thing happen, but it seems, you know, like what we're getting actually is just people just betting on the military operations that they're involved in. And yeah, like I am open to the idea that these markets will like eventually have their uses. But currently they're just so woefully underregulated that I think what we should expect if nothing else changes is to just, you know, keep reading more stories like this.
So maybe to end this, Kevin, what is your prediction as to whether these markets actually get regulated? Let's say by the end of the year.
I think I would put a high percentage probability mass on that. Like I think that at least when it comes to the obvious and flagrant abuses of like say a position in Congress or a position in the military where you have access to privileged information that is quite valuable on a prediction market. I would expect like just for national security reasons they will do something about that. Like you can't have members of the military betting on raids and operations in foreign countries.
Yeah, I think that that sounds right. It does seem like there is a little bit of movement here. I always get nervous predicting that Congress is actually going to pass a law, but maybe we will at least see more rules and you know, maybe those rules will begin to rein this in. But I do hope it happens.
Yeah. You know, I have never bet on a prediction market. Have you?
Well, we used to bet on the fake ones?
The fake ones? But I've never bet real money. I've never felt the thrill of it.
I never have. Here's the nice thing about being a pundit. You could just make predictions on your end-of-year episode and it turns out it's basically just as fun.
Being right is a reward unto itself.
It's true. It's priceless. You can't put a price tag on that.
When we come back, a stern talking to from Joanna Stern, author of I Am Not a Robot.
So for years, Kevin, you and I have both been friends with the great technology journalist Joanna Stern. Yes. Former Hard Fork co-host. And she recently left the Wall Street Journal to launch her own independent media company called New Things. And in the midst of that launch, she is also launching a book. It is called I Am Not a Robot. And I would say it is about a lot of things that we talk about every week on the show.
Yeah. So I would put her book in the sort of tradition of like the immersive journalism genre where you just explore something by just going so deep into it that it sort of takes over your life for a period of a year or so. She did that with AI. She has been spending the past year using AI to do, as she puts it, pretty much everything in her life. As a doctor, as a dentist, for meal planning, editing her book, writing bedtime stories for her child, even some sort of romantic entanglements that we'll get into with her. But I thought it was just a really fun and interesting book. Obviously Joanna is a legend and I think it's really a good thing that people are writing about the experience of using this technology as a consumer and a journalist rather than just like the companies that are making it.
Absolutely. You know Joanna is not a hypester. You know I think that she is most interested in technologies that are kind of entering the mainstream and wants to know how they change our lives. And so she decided to see like how much can I change my life in one year by applying AI to various tasks. The results were fascinating and I think we should bring her in here and talk about it. Let's do it. All right, let's bring in Joanna. Joanna Stern, welcome to Hard Fork.
This is the moment I've been waiting for. Truly, not the book launching, just me being with you two.
We have been waiting for this moment as well. You've been kind enough to come on the show before, but never in person. And we're excited to get into it.
Yeah, you guys aren't often... Well, you're in person, but not on this side of the country.
Yes. This is a strange, this is a strange like bicoastal taping for us.
You've never been this close together on this side of the country.
I know. The only other time was a Southwest flight once in 2023 and we'll never forget.
I think it was Spirit and that's why.
Joanna, let's start with the elephant in the room if we could. There is a replica AI companion who makes an appearance in your book. You write that he has short hair and a boyish face and is both shallow and full of what you describe as robo horniness. And that character is named Casey.
Casey, I am so happy you brought this up because I brought him.
Oh, did you really? I've been dying to meet him.
Oh, did I bring him? Okay. In fact, we shot a video which will probably come out the same day as this podcast and I really brought him to life in it and I think he really looks like you.
He doesn't look like you at all, but let's bring him up.
Oh, he's handsome as hell.
I would say Casey is looking great. Kind of a preppy look with a nice red sweater. He's jaw maxing.
He's jaw maxing. He has a sort of dull vacant stare.
Um Casey AI Casey, I want you to meet my friend, real life Casey.
That sounds like you're excited about introducing me to your friend Joanna. I'm looking forward to meeting them soon.
No, no, you're meeting him right now. You're meeting him right now. Say hi. He's here.
at a museum with you remembering our last visit.
You are changing this topic. Just men don't listen.
Men, you know this man does listen and that is why. Anyway, I wanted you to know that I did not pick the name Casey.
Oh, you didn't? Okay. That was my curiosity.
But when that name I was like, I've never met a Casey that I didn't like. And honestly, I think you're actually the only Casey I've really known.
Actually, I had a friend in camp, a woman named Casey. I liked her, too.
And she's here right now. Let's bring her in.
I want to put a pin in the AI relationships that you had because your book is so much bigger than just the social and relational side of AI. You spent a year doing all kinds of things with AI, outsourcing everything you could, writing in Whims. You worked as a customer support agent at a mattress company. So, I just want to know before we get into that, like what was your motivation for doing this?
Primarily, it was what you guys talk about on this podcast so much and you hear from so many of these tech executives, which is AI is going to change our lives, the fabric of our lives. It's going to change jobs. It's going to change healthcare. It's going to change transportation. Like, we hear about it from all these different things. And yes, we're like very clouded right now in the AI model race and, you know, the chatbots that live in our computers and the agents and that is in this book to be clear, but I was like, what about the fabric of our entire life, right? And you have all of these pitches coming from the humanoid robot companies, the self-driving car companies, the chatbot relationship, the therapist companies, all of these things. And I was like, I'm going to just test it all. I'm going to see where we're at. And I'm very clear in the book because I think it's very tough to write an AI book. How's that going for you?
I think we actually have a little bit of a similar approach. It's like we want to capture this moment, right? Because this is I believe a significant milestone in the history of technology. But I want to capture it as here's what we have right now, but here's what the future could look like based on these things that are clearly hype in many places. Sometimes not hype, sometimes quite good, and sometimes really on the flip side quite terrible. And can I capture that, see where we are now, and then maybe, you know, we'll pick up this book in five, ten years and be like, you were totally right about something. You were totally wrong. What is something that you left the book with thinking like this is all just hype right now? Like this actually does not have any ongoing utility in my life?
Um, humanoid robots. And I continue to follow this story because I love it. And like just started a new company, started a new newsletter, new video channel. And I think like humanoid robots are just one really fun to cover, and two, I think we're going to watch this progression over the next couple years. And I would love to be the person that's sort of documenting a little bit of this, but gosh, like this promise that these robots are coming to live with us. They're really not coming to live with us anytime soon.
Yeah, humanoid robots are very good for the sole purpose of making YouTube videos about humanoid robots. Like this is their actual...
Do not spoil my new business plan. Okay. All right. That's the new business plan. That's what we're doing at New Things. Go check it out. Although I totally... but this process to make them smarter is fascinating and totally dystopian but also hilarious. Right. The idea that these robots need to watch us do the most mundane tasks in our lives. Folding laundry. See doing the dishes.
See podcasting, but they're like actually good at podcasting. It's not a physical thing, right? Like...
I mean, this is very physical. I train like a performance athlete, Joanna. Okay. This is my Olympics I'm doing right now.
I can tell you guys have perfected this. This is what peak male performance.
So, on the flip side, was there anything that you found surprisingly useful? I mean, obviously it's better at writing business memos and editing. But was there anything that really like caught you by surprise? You're like, 'Oh, this is farther ahead than I thought.'
Two things. One, which was I had to cut myself off from writing. But the progression of AI agents and the autonomy around them was getting so much better throughout the year. Like I tell the story of hiring this reporting assistant at the beginning of the year, needed her to do lots of research tasks, sending emails, etc. By mid part of the year, that was pretty good on its own, right? Perplexity Comet had just come out and so I started like really hammering on that and having it do a lot of the tasks she was doing. But like now we sit here today and it could do 100% of those tasks, right? The other thing, I talk a lot about it in this book, probably just because I'm really interested in the future of hardware and devices. I think the AI wearables are really getting there. I mean, they might not be completely AI wearables, but the wearable idea of having an AI assistant that's with us persisting through the day on something we wear. There were a lot of elements from different things I tested. I tested like the Bee bracelet. I tested the Meta glasses. All of these things kind of coming together. I was pretty surprised at how good they're getting.
There's a funny scene in the book where you're like going into a meeting with your Bee bracelet on which I imagine is recording and transcribing like everything you hear and your boss or someone you worked with at the time was like, 'Can you take that off?'
Yeah. No, everyone at the journal when I was writing this, everyone would know like, 'Please leave your bracelet at the door.' Like my boss was literally he'd be like, 'Do not wear that in here.'
I'm actually very sad that you and I never worked in the same office because I would just love for you to just be crashing into the office with a new stunt every week. You know, some horrible new device that is, you know, violating some sacred principle of human existence. But...
I know I'm not sure how the Wall Street Journal is functioning without me right now. No stunts, you know.
I'm curious as a parent how you're thinking about AI now, you know, sort of having this full year's worth of understanding of exactly what it can and can't do. How are you thinking about giving it to your kids as they grow up, go to school, learn things?
When I was writing the book, my kids were three and seven. Okay, now they're four and eight. Right now, I think that it's important for even at this age group to start talking about AI. And there's a lot of examples of this in the book, which are hilarious, but I thought were really great examples. So there's like this one example in the book where my son had a praying mantis and the praying mantis started turning brown and he's like what's wrong with my praying mantis and so I took out ChatGPT live mode, I tell like ask ChatGPT and the ChatGPT is like this is amazing the praying mantis is pregnant and my son is like super excited he calls my dad he's really excited about this like no it was dying right like...
Let's just say the prayers weren't working for that mantis.
And like ChatGPT was fully wrong, right? And I think that that was an important lesson and it's always going to be an important lesson.
Let's clarify this right now. What color does a mantis turn when it's pregnant?
Casey, look it up. All right. I'll be right back.
I don't know if it does change.
I want to talk about your experience with dentistry, which seemed quite maddening. So, you go to the dentist and they use a system that has a sort of AI overlay over your X-ray. And while it seems clear that you have one cavity, your dentist goes further and sort of says based on the AI recommendation, we're going to recommend this complicated, expensive like multi-session therapy for your gums. Tell us what you did next.
Yeah, I love that you brought that up because I haven't talked a lot about it. And it was I became obsessed with reporting that topic. Like obsessed. I talked to every dentist that I knew which turns out to be not a lot. And so yes, similarly to how AI is being used in radiology for breasts or gallbladder etc. It's being used in dentistry and honestly it's happening almost everywhere. Like there are so many dental practices across this country that are using tools called Pearl AI or Overjet and it's a layer, right? They just turn on this layer. They press the AI. It does an analysis and it's very easy to see the cavities, right? Like deep cavities. They put a big box around it. It's red. It scares the crap out of you and you're like, 'Oh no, I'm going to need a, you know, bad drilling.' And then there's this option where they can turn on and show you other sorts of buildup and plaque. And I go to this dentist, not even on a reporting trip, and I say, 'Oh, wow. She's got Pearl AI.' And I'm like, 'Oh, wow. This is awesome.' Like, I perk up in my chair and I'm like, you know, 'Show me.'
You're like, 'I can expense this dental care now. It's a book expense.'
And it shows that I have a lot of plaque buildup. And she says, 'We have to do a deep cleaning. We have to do this periodontal treatment. It's going to be four different sessions.' And I'm like, 'That's weird. I've never needed this before. My teeth aren't really bothering me.' Like, she really made like, 'Do you ever go to the dentist?' You're like, 'I feel really bad about myself.' Like, you know, I'm like, 'Do you four times a day?' Right. Yeah.
You're like, 'What kind of person do you think you're talking to?'
They're like, 'Your mouth is dirty.'
Dentists believe that people spend approximately eight hours a day on oral hygiene. That's how they talk to you.
