How Predictions Took Over Our Lives  🔗 **Source:** [How Predictions Took Over Our Lives](https://share.snipd.com/episode/ff651f94-5b5d-4627-8798-cdfc51f6adc9) ## Highlights - **Students Bet On Corner Kicks Not Winners** - Brendan Dwyer watched students betting on the number of corner kicks during a Morocco vs Belgium World Cup match instead of the match winner. - The students cheered or booed every referee call because their bets depended on corner kicks, and they remembered that detail more than the final 2-0 result. Transcript: Brendan Dwyer Was the FIFA World Cup in Doha, Qatar. So it was in November, which means my students were here. They were in our workroom and they were huddled around the laptop. Manoush Zomorodi Brendan Dwyer is a former football coach and now a professor of sports leadership. He's at Virginia Commonwealth University, which is a financial supporter of NPR. On this day, his students had bet on a game, but not the way Brendan expected. Brendan Dwyer They were watching Morocco and Belgium, which is a very unconventional game for these students to watch because they weren't from Morocco or Belgium. But these students were like enthralled by this game. And as the game was going on, more students were starting to gather around. And it was not a knockout round game. It was really not a semifinal or a final. And it was an interesting game, but we're talking second half. It's not the last couple minutes. But each moment, they were yelling and screaming and getting upset. I found that they were getting upset mainly at the referee, more than they were getting upset at, you know, a missed shot or a missed pass or a missed play by one of the players. Okay. That's when I was like, what are you guys cheering about? And they mentioned that they weren't betting on the outcome of the game. They were betting on the number of corner kicks. Manoush Zomorodi In other words, they weren't cheering for either team specifically. They were just hoping to see more corner kicks because they put money on it. Brendan Dwyer So every time the referee awarded a corner kick or did not, they reacted a certain way. And that was hilarious to me and also at the same time got me a little caught off guard because I've never seen, one, I didn't realize you could bet on the number of corner kicks at that point. And two, the fact that these students were so enthralled, not by who's going to win a World Cup game, but corner kicks. And the students won. They picked the over and they won the over. What ([Time 0:01:04](https://share.snipd.com/snip/a2fda2bc-4754-49a3-b4e1-642a51d154cf)) - **2018 Ruling Fueled Massive Online Betting Growth** - The 2018 Supreme Court decision that allowed states to legalize sports betting exploded the market from billions to roughly $150 billion by 2024. - Brendan Dwyer notes 95% of that is online, fueling new viewing habits and economic incentives across sports. Transcript: Manoush Zomorodi TED stage. A brief history. Brendan Dwyer So in 2018, the Supreme Court struck down the federal ban on sports gambling. Since then, 37 states plus the District of Columbia passed laws where we can bet on sports. Manoush Zomorodi Since he gave this talk in 2025, two more states have joined that list. Brendan Dwyer That has been huge for the betting industry. The American Gaming Association estimated in 2017, just under $5 billion was spent on sports betting, primarily physically in a casino. In 2024, that number was $150 billion, 95% of it online. The sports betting industry has ([Time 0:07:08](https://share.snipd.com/snip/ac774a5e-cfcb-456c-8e36-f284c0b282a5)) - **Betting Turns Fans Into Day Traders** - Sports betting transforms fans from passive viewers into over-engaged spectators who act like day traders tracking open bets. - Brendan Dwyer found highly involved bettors watch more, consume more analysis, and treat the scoreboard as a balance of wagers rather than meaning. Transcript: Brendan Dwyer Well, I mean, there are all sorts of different sports betters. They consume at all different levels. As your involvement, which is a attitude, not necessarily a behavior as your cognitive involvement. Your attachment, and then your emotional involvement increases, the more you consume everything. And so we see sports bettors from the lowest level that are consuming more than non-betters, but then those that are at the highest level, the ones that consider themselves almost predictive Analysts are consuming so much sport because they want to find that edge. Sports bettors are not disengaged, they're over engaged. And those highly involved sports bettors, they are part spectator, part day trader. The scoreboard is no longer where you put the final result. It's a balance of open bets. Now we have games sometimes starting at 9 a.m. And ending at 2 a.m. That means the book is always open, creating a perpetual loop of deposit, wager, outcome, repeat. And when loyalty, attention, and money become so intertwined, so do the incentives. At every level, the league, the team, fans, even the athletes. The question isn't whether sports betting enhances engagement. It clearly does. The question is, what type of engagement is it? My students would later tell me they remember the corner kicks. They don't remember who won the game. It was Morocco, by the way, 2-0. And that is the trade-off. ([Time 0:15:20](https://share.snipd.com/snip/f2b1d69e-9a5a-42ba-ad60-ec2f43a67d71)) - **Young Bettors Have Illusion Of Control** - Younger bettors (18–24) combine overconfidence and impulsivity, viewing betting as a money-making opportunity despite risks. - Dwyer's research shows youth perceive skill in sports betting, increasing an illusion of control that can be harmful. Transcript: Brendan Dwyer I guess I could start out by saying pretty much any sports fan thinks they know more about sports than everyone else. So I think, you know, especially when you talk about 18 to 24 year old males, they have a built in level of confidence all about a lot of things, but in particular about sports, it is also At a time in their life where for a lot of them, money is important to them. And they oftentimes do not have a lot of it. From what we found in our research, they see betting as an opportunity to make money. And it's that combination of overconfidence and impulsivity of younger individuals that is troublesome. That's really more evident in younger people than older people. And older people we found in our research are more rigid, which also comes with challenges in terms of changing their perceptions and their behaviors towards betting. But I think majority of sports bettors are very responsible bettors. Oh, really? Manoush Zomorodi They are? Brendan Dwyer I think they're irrational. But I think in terms of the amount of money they spend, they have it under control for the most part. But the younger individuals, overconfidence and impulsivity, if their belief in sports betting is an opportunity to make money, is really a challenge. Manoush Zomorodi Do the words matter? Like if you said to a young person who's into sports betting, you know you're gambling, would they push back on that? Would they say absolutely not? Brendan Dwyer No, I think they enjoy the risk. Another study we have conducted found that risk plays no role in their decision to gamble. If they know it's more risky, that's not going to make any impact on their decision. I work with 18 to 24 year olds on a daily basis. And I worry about their decision making for anything they do. They have overconfidence and regulation issues to everything. And so when you add overconfidence and sports and money, it worries me even more because it's not just like going to Vegas. Games, there is some skill in luck. You know, if you play blackjack, if you play poker, craps, there's more skill in those games. And if you play roulette or a slot, those are more luck based. In sports, you can have a much higher perception of the skill. ([Time 0:18:27](https://share.snipd.com/snip/5fb3dc04-2b43-4c9f-b361-c20e5f1cd744)) - **AI Predictions Are Social Magnets Not Neutral Facts** - Machine learning is fundamentally a prediction engine that projects past data onto new cases, which makes social predictions powerful and consequential. - Carissa Véliz highlights that predictions about people bend reality toward themselves and are not neutral facts. Transcript: Carissa Véliz What it means is that it uses data from the past and projects it onto data it doesn't have. And I think that we haven't fleshed out the implications of that. Because if predictions were only about knowledge, then, fine, this is a scientific endeavor. But they're not. And one very basic reason for why they're not is that predictions are never facts. They can be educated guesses. They can be power plays. They can be wishful thinking. They can be estimates. But the future is unwritten and facts belong to the past. Even though we tend to associate prediction with knowledge, I'd like to invite you to consider the possibility that most of the predictions that you encounter in an everyday setting Are closer to the realm of power than that of knowledge. Carissa Villiz continues from the TED stage. It might seem that the days in which we sought astrologers and soothsayers to tell us about the future are very distant. But often we use AI as the new oracle of Delphi, and tech executives whisper in the ears of our leaders, much like court astrologers used to. Granted, the technology is different, but the political role is not. Predictions are often power plays in disguise. They justify value-laden decisions under the pretense of facts. Better understanding prediction matters more than ever because we're relying on forecasting more than ever with AI. And based on how we talk about prediction, we're being much too naive about it. But AI is science, you might think. It's cutting-edge technology, right? Well, it depends on the kind of AI and how we use it. AI can be a great technology to make predictions about molecules in the search for new antibiotics. But predictions about human beings are fundamentally different than those about things. Predictions about the weather don't influence the weather. Predictions about people influence people. Social predictions tend to act like magnets. They bend reality towards themselves. They affect reality they purport to predict. ([Time 0:25:42](https://share.snipd.com/snip/f6153197-fb33-40d0-9a0b-f34ce9abb4fe)) - **Question Claims That AI Futures Are Inevitable** - Carissa Véliz urges skepticism when tech executives present AI forecasts as inevitable futures and to question their incentives. - She advises treating such pronouncements as marketing ploys that can create self-fulfilling demand rather than facts. Transcript: Carissa Véliz They justify value-laden decisions under the pretense of facts. Better understanding prediction matters more than ever because we're relying on forecasting more than ever with AI. And based on how we talk about prediction, we're being much too naive about it. But AI is science, you might think. It's cutting-edge technology, right? Well, it depends on the kind of AI and how we use it. AI can be a great technology to make predictions about molecules in the search for new antibiotics. But predictions about human beings are fundamentally different than those about things. Predictions about the weather don't influence the weather. Predictions about people influence people. Social predictions tend to act like magnets. They bend reality towards themselves. They affect reality they purport to predict. Manoush Zomorodi So let's explain a little bit more, just if people are uncertain about how AI is built on prediction, in that these models have digested basically the entire Internet. To what people have written previously, how they've constructed sentences, words, ideas, and then simply replicating that when we ask them a query, when it comes to LLMs at least. Carissa Véliz That's exactly right. The large language model outputs a likely response. And many people think that it's a response that's likely to be true, but that is actually imprecise. It's a response that is likely to have been given by a person, had you asked a person, and likely to have been accepted by another person. Because the way these models were trained was with human beings. So they would suggest two answers to a human being, and then a human being would prefer one of those, and then the model would learn what human beings prefer. But as we well know, sometimes human beings don't prefer the truth. We prefer good stories. And so the models are not trained to track truth. Manoush Zomorodi Can you tell us more about some of the use case scenarios you're seeing where AI is being used as a forecasting tool in terms of trying to predict human behavior that you think is dangerous? Carissa Véliz Yeah. So insurance is one very important one. And we're going from population level predictions, as in how many people in 10 million are going to get diabetes, to individualized predictions. Is this person going to get diabetes? And that is much more dangerous because it creates self-fulfilling prophecies much more easily. Because for example, if I predict that a person is going to have bad health and then I put his insurance premium up, that in itself might lead to more stress, which leads to worse health. So I affect that person very directly. Human beings are so suggestible and we can be so stressed out by circumstances that corner us. And the whole point of insurance is to pull risks such that the unlucky don't have to burden their unluckiness alone. So another use case is giving out any kind of opportunity. So using AI for giving out loans, for giving out jobs, for deciding whether somebody should have an apartment, for making decisions in the justice system having to do with bail or with Sentencing. And as far as I can tell, we are using AI pretty much in every sphere of life. Manoush Zomorodi The argument that I hear with many of these examples actually is, well, more information is good information. And so why would we not look to the past and look at trend lines and the data so that we can make better choices going forward? Carissa Véliz Essentially, we are reducing people's capacity to defy their odds. If you take into account that predictions are not facts, arguably they don't count as information. The second point is that when you're looking for a needle in a haystack, if you add hay to the haystack, it doesn't help. So no, more information is not always relevant and it's not always helpful. And the third thing is it depends on how we use that data. Manoush Zomorodi Do you think people say that there are ways to check for algorithmic bias and make sure that the modeling gives us decisions that reflect a world that we want to live in, not necessarily One that we have lived in? Carissa Véliz With an algorithm that is not biased. If I deny you an opportunity on the basis of a clear criteria, either it's a loan or a crime or bail or something. If I deny you a loan because you don't have X amount of dollars in your bank account, or I deny you bail because you committed X crime, those are contestable facts. We can argue about them. If I'm wrong, you can challenge me. But if I deny you an opportunity on the basis of a prediction, how are you going to challenge that? Because a prediction is not a fact. So how are you going to prove to me that you will pay the loan or that you are safe? You can't. And so when we use predictions to make important decisions, what we are actually doing is creating literally a Kafkaesque world in which we are ruled not by principles, not by criteria That are transparent, contestable, not by due process, but by statistical pattern matching against which there is no argument. Manoush Zomorodi In a minute, Carissa Valiz weighs in on whether our AI future really is inevitable. On the show today, predicting the future. I'm Manoush Zomorodi, and you're listening to the TED Radio Hour from NPR. We'll be right back. Simone Stolzoff This message comes from Schwab. With the new Schwab Teen Investor Account, teens can gain hands-on investing experience. It's co-owned by you and your teen, so you can monitor the account while your teen learns how to invest and manage money. Learn more at schwab.com. Manoush Zomorodi Support for this podcast and the following message come from ActiveCampaign, the autonomous marketing platform. You know that feeling when you open your marketing tool and instead of marketing, you spend an hour wrestling with a drag-and builder? ActiveCampaign built active intelligence for exactly that moment. Describe what you want to accomplish. It builds the campaign, writes the copy, and maps the automations across email, SMS, and WhatsApp. Customers save an average of 10 hours per week and make email campaigns eight times faster. Learn ([Time 0:26:59](https://share.snipd.com/snip/6f2d78b7-5437-42c2-a0da-caf08fd6ceb5)) - **Individualized AI Decisions Create Hard To Challenge Outcomes** - Using AI for individualized decisions (insurance, loans, hiring, bail) risks creating self-fulfilling prophecies and unchallengeable outcomes. - Véliz warns predictions used as opaque decision tools produce Kafkaesque systems with little recourse for affected people. Transcript: Carissa Véliz Yeah. So insurance is one very important one. And we're going from population level predictions, as in how many people in 10 million are going to get diabetes, to individualized predictions. Is this person going to get diabetes? And that is much more dangerous because it creates self-fulfilling prophecies much more easily. Because for example, if I predict that a person is going to have bad health and then I put his insurance premium up, that in itself might lead to more stress, which leads to worse health. So I affect that person very directly. Human beings are so suggestible and we can be so stressed out by circumstances that corner us. And the whole point of insurance is to pull risks such that the unlucky don't have to burden their unluckiness alone. So another use case is giving out any kind of opportunity. So using AI for giving out loans, for giving out jobs, for deciding whether somebody should have an apartment, for making decisions in the justice system having to do with bail or with Sentencing. And as far as I can tell, we are using AI pretty much in every sphere of life. Manoush Zomorodi The argument that I hear with many of these examples actually is, well, more information is good information. And so why would we not look to the past and look at trend lines and the data so that we can make better choices going forward? Carissa Véliz Essentially, we are reducing people's capacity to defy their odds. If you take into account that predictions are not facts, arguably they don't count as information. The second point is that when you're looking for a needle in a haystack, if you add hay to the haystack, it doesn't help. So no, more information is not always relevant and it's not always helpful. And the third thing is it depends on how we use that data. Manoush Zomorodi Do you think people say that there are ways to check for algorithmic bias and make sure that the modeling gives us decisions that reflect a world that we want to live in, not necessarily One that we have lived in? Carissa Véliz With an algorithm that is not biased. If I deny you an opportunity on the basis of a clear criteria, either it's a loan or a crime or bail or something. If I deny you a loan because you don't have X amount of dollars in your bank account, or I deny you bail because you committed X crime, those are contestable facts. We can argue about them. If I'm wrong, you can challenge me. But if I deny you an opportunity on the basis of a prediction, how are you going to challenge that? Because a prediction is not a fact. So how are you going to prove to me that you will pay the loan or that you are safe? You can't. ([Time 0:29:15](https://share.snipd.com/snip/d4423008-1d39-4c02-85b5-fad5cf589aeb)) - **The Web Is Built To Predict And Influence You** - The infrastructure of the internet exists to power prediction: data is collected to forecast and influence future behavior, trading autonomy for predictability. - Véliz argues this shapes what we see and do online, making certain behaviors easier to predict and thereby reducing autonomy. Transcript: Carissa Véliz So we only collect all of this data and watch people because we're trying to figure out what they're going to do next. And we're trying to influence that. And that's not that different from being in the savannah thousands of years ago and have a lion looking at you because it wants to know where you're going to go next. So it can pounce on you and eat you. It's the same principle. Manoush Zomorodi But some people would say, well, okay, so I log into Amazon and it knows that I need more detergent. I'm grateful for that prediction. Do you find something dangerous or sinister in that? Carissa Véliz Well, no, if it's only that, but it's never going to be only that. And part of why predictions online are true is because it's easier to determine the future than to predict it. So, for example, to give an extreme example, just to illustrate the point, if I want to be absolutely sure of where somebody is going to be tomorrow, the easiest thing to do is to jail them. Because that way, I will not figure where they're going to be. So when the online world influences our views in a way that makes it easier to predict our behavior, we are paying a cost in autonomy. ([Time 0:39:39](https://share.snipd.com/snip/067704db-3339-42cb-a1d2-6cfec575b5aa)) - **Man Lets Algorithms Choose Where He Lives** - Max Hawkins used algorithms to inject randomness: an app sent him to random San Francisco locations and later chose cities for him to live in for months. - This revealed overlooked local scenes but eventually felt like avoidance rather than true freedom. Transcript: Simone Stolzoff And I would drink my coffee and get onto my bicycle and ride into work. And I had, I optimized my schedule to be perfect. And so in order to get outside of his bubble, outside of his comfort zone, he did something rather extreme, which is he started designing different algorithms to infuse randomness In his life. The first thing that he did was he designed an algorithm that would call an Uber to his house and bring him to a random location in the city that he didn't know. And I was hooked. I started using this app to go to all different places in San Francisco. I went to museums randomly, random grocery stores, random bars, random bowling alleys, random florists. And I started discovering that there was an entire side to San Francisco that I had been ignoring. Of my preference. And at the most extreme example, he quit his job and designed an algorithm to choose where he lived. And it sent me all over the world. Taipei, Taiwan, Mumbai, India, Dubai. And every time I would go to a new city, I would do the same sort of stuff I was doing in San Francisco, go to random events, meet random people. And I'd live there for two to three months and then ask the computer again for the next location. I did this for two years. ([Time 0:47:11](https://share.snipd.com/snip/2c5d2486-3fc2-4605-9e43-b2217949e6f2)) - **Microdose Uncertainty To Build Tolerance** - Microdose uncertainty to build tolerance: take a different route, try a new restaurant, or inject small, controlled novelty into routines. - Simone Stolzoff suggests these low-risk experiments increase comfort with ambiguity and open creative possibilities. Transcript: Simone Stolzoff Maybe you take a little different route to work or you go to a restaurant that you've never been to. And through exposing yourself to that novelty, you not only discover more about yourself, but you're training yourself to be able to be more tolerant for uncertainty in the future. Manoush Zomorodi So what role does prediction, healthy prediction, have your view? Like how do we balance the very real possibilities with this inability to ever really know what's going to happen? Simone Stolzoff So rather than having very fixed goals or having a certain idea of my 10-year plan, I try to treat my life a little bit more like a prototype, like an experiment, and try to approach the Uncertainty in my life with curiosity instead of fear. There's this idea from Buddhism that I really like, which is to focus on the next right action. When we're faced with uncertainty, I think it's really easy to loop through all of the worst-case scenarios or all of the logistical and emotional ramifications if things don't go Our way. But rather than worry about all of those next steps, how might you try to stay grounded in your sphere of influence, of what you can control right now? It's incredibly disempowering to have a fixed idea of exactly how things will go, because that robs you of the ability to shape it. ([Time 0:49:54](https://share.snipd.com/snip/051379ac-e749-4e19-8b1b-a72440d91adb))