Pod Insights Hard Fork BONUS: Hard Fork Live! Big Tech’s Arch Nemesis + Bot, or Not? About this episode Jonathan Kanter, who heads up the Justice Department’s antitrust division, believes that antitrust laws are critical for innovation — from ad tech to A. I. The assistant attorney general is bringing a new philosophy to enforcing those laws. So, how is his new approach to protecting competition playing out?
Plus: Can you guess whether that was a bot, or not?
Transcript Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.
0:00 Speaker 00 Your agents can talk to each other. They still can't think together. That's the gap Outshift by Cisco is closing. Outshift is Cisco's incubation engine, building the internet of cognition, shared intent, shared memory, and guardrails that allow agents to work together from anywhere. Open source protocols and code, a system for intelligence that scales. Read the paper and experience the demo at outshift.com. That's O-U-T-S-H-I-F-T.com.
0:32 Casey Newton? Kevin, what are we doing here? It's not Hard Fork Friday.
0:34 Kevin Roose? It's not Hard Fork Friday, and yet, we have a new hard fork.
0:37 Casey Newton? That's very exciting.
0:37 Kevin Roose? Yeah, this is a special bonus episode. So, a couple of weeks ago, we told you that we were doing a show at South by Southwest, Hard Fork Live, and we wanted to share that with you today.
0:48 Casey Newton? I had a great time.
0:49 Kevin Roose? Yeah, it was really fun to go on the road. Thank you to everyone who made it out and stuck around afterwards to say hi, take selfies, et cetera. Very fun.
0:58 Casey Newton? My favorite listener, and I will not sort of give away any identifying personal information for reasons that will soon become obvious, but, you know, in order to see the Hard Fork Live show, you were supposed to have a South by Southwest badge. And this person somehow talked their way past the security guard and just said, "I really want to see the show," and was able to see the show illegally at a cost to South by Southwest of several hundred dollars. So, we just wanted to say thank you and respect to the people who would do anything to see the show.
1:28 Kevin Roose? Yeah, shout out to all our podcast pirates out there.
1:30 Casey Newton? Mm-hmm. One thing we think you would love to hear is our conversation with Jonathan Kanter.
1:36 Kevin Roose? Yeah, Jonathan Kanter is the Assistant Attorney General of the Department of Justice. He is responsible for going after the big tech companies for antitrust violations. And he's been sort of the top cop who's been really scrutinizing and pursuing charges against these big tech companies.
1:53 Casey Newton? And, you know, since Joe Biden became president, the antitrust division of the Department of Justice has gotten much more active. They have opened investigations into what feels like most of the biggest tech platforms. And that makes him a really interesting person as we think about what the future of the tech industry is going to look like.
2:10 Kevin Roose? Totally. And so, we talked with him. You'll hear that conversation. And then, we played a fun little game with our live South by Southwest audience called Bot or Not. And you can hear that as well.
2:21 Casey Newton? And play along at home, if you like.
2:22 Kevin Roose? Yeah.
2:22 Casey Newton? And email Kevin your score.
2:24 Kevin Roose? Don't do that.
2:25 Casey Newton? All right.
2:31 Kevin Roose? So here, live from Austin, is the first ever Hard Fork live.
2:35 Casey Newton? And if we could just sort of pipe in the sound that the audience made when the Beatles performed on the Ed Sullivan Show in the 60s, that's how I want it to sound.
2:43 Kevin Roose? Thunderous, roaring, hysterical, cheering.
2:46 Casey Newton? Freaking teenage girls.
2:48 Kevin Roose? It's weird. Why does our, why does our live audience in Austin keep saying Ringo? What's that? What's that about? Anyway, here's our show.
3:05 Speaker 03 Building and scaling AI today can feel pretty chaotic, mostly because teams are juggling too many fragmented tools at once. That's where OutSystems comes in. OutSystems gives you one unified agentic systems platform to manage your entire portfolio of agents, apps, and workflows. You get to innovate at the speed of AI, but with the security, governance, and control your enterprise actually needs. And the results are already there. A major manufacturer used OutSystems to build a multi-agent system that analyzes blueprints, and a global bank cut their development time by 60%. Own your agentic future today at outsystems.com/NYT.
3:42 Speaker 04 Your sales team is at risk of missing quota. Don't panic. Just ask Rippling AI. Built on your real-time people and business data, Rippling AI can pull metrics from Rippling and your CRM into a meeting-ready dashboard showing quota attainment, headcount trajectory, and revenue to quota by region. In seconds, you'll see what's behind your quota risk and fix it before it's missed. Head to rippling.ai/hardfork to get the only AI built to give you full visibility and take complex actions across your whole business. That's R-I-P-P-L-I-N-G. dot AI slash hardfork. Sign up today.
4:16 Speaker 03 Stripe is built for whatever comes next. With Stripe Billing, you can easily test and iterate subscriptions, launch new usage-based and hybrid pricing models, and automate invoicing. Stripe Billing supports whatever business model you want to launch next. And Stripe powers all the Forbes AI 50 who have monetized. Learn how you can power any business model at stripe.com/billing.
4:40 Casey Newton? Please welcome Assistant Attorney General Jonathan Cantor of the Justice Department.
4:50 Kevin Roose? Hello.
4:51 Speaker 05 Hello. Great to see you.
4:52 Kevin Roose? Welcome.
4:53 Speaker 05 Thanks for having me.
4:54 Kevin Roose? Thanks for coming to Hard Fork.
4:55 Speaker 05 Yep.
4:56 Kevin Roose? I want to start by reading a couple of the monikers that the press has given you over the years, having you react to them. So you've been called, quote, "top antitrust cop", "Google foe", "Big Tech's arch-nemesis", and "an anti-Google legal warrior", that one was by The New York Times. How does it feel to be referred to as something like a Marvel villain?
5:23 Speaker 05 I was gonna say hero.
5:26 Speaker 05 And did you get that from a chatbot?
5:31 Kevin Roose? Well, speaking of chatbots, I heard that you consulted one in picking out your outfit today.
5:35 Speaker 05 I did, so I've got colorful shoes on, as folks here might be able to see. And I asked ChatGPT, you know, "What do people think of colorful shoes?" And the response was, "Well, it really depends on the context, but if it's a music or art festival, colorful shoes are okay."
5:54 Kevin Roose? Yeah.
5:54 Speaker 05 So here we are. I will note, though, that I was gonna wear them either way.
6:00 Kevin Roose? Well, you are dressed down for a regulator, which we appreciate here at South by Southwest.
6:07 Kevin Roose? I want to talk about some of the tech companies that the Department of Justice has investigated. But first, I just want to kind of lay some groundwork for this conversation and talk about you and your background and your sort of philosophy of antitrust, and how you got to the place you are. So you were, before you were at the Department of Justice, you were a lawyer in private practice, working with and for some of the big tech companies. How did you develop your theory of tech antitrust?
6:37 Speaker 05 Sure, so my experience informs my outlook. And I, over time, learned that when we apply antitrust law, there are these very formalized boxes, and we try to fit everything into this narrow framework, assuming that, for example, consumers will behave rationally, and they'll all behave the same way, and they'll run from one store to the other. And then, as I got more into tech antitrust, what I realized is that the world has changed in ways that are, over the last 20 years, that are on par with, if not exceed, the Industrial Revolution. Market realities have changed fundamentally. Think about your world today versus 20 years ago,
7:21 Speaker 05 for those of you who are alive 20 years ago. But it's changed dramatically in the way we conduct business, in the way we gather information, and in the way we behave as individuals. And so we were applying this old framework that was billed for a world of poles and wires to a world of ones and zeros. And so from my perspective, we needed to update our framework to reflect those market realities.
7:47 Casey Newton? Well, I mean, let's talk in a little bit more detail, right? I think the thing that comes up the most is that in the past, the main way that antitrust would be enforced is if prices rose for consumers or there was some sort of like obvious financial harm to consumers. But now we live in a world where most of these services are free to use, or certainly the ones that regulators have been most interested in, like Google and Facebook. And so at the same time, there's been this growing sense that these companies are awfully big and it seems like there's some anti-competitive behavior that they're engaged in to sort of stay as big as they are. But they're free. And like, anybody could use anything else if they wanted to. So there sort of had to be a new theory of the case with regard to antitrust. Is that the sort of new market reality you're talking about?
8:29 Speaker 05 In part, yeah. I mean, ask yourself, is anything really for free? But so what are you giving up to get something? Are you giving up your privacy? Are you giving up your data? Are you giving up your personal profile that can be used to create in the future an AI simulation of yourself? Those are the kinds of questions we need to start answering. The other point, I think, is really important because when we think about price is when we think about antitrust just in the context of price, one problem is that we're missing the point of antitrust enforcement and anti-monopoly law. So antitrust principles have been embedded in our society since our founding. The Tea Party,
9:14 Speaker 05 the Boston Tea Party, was an anti-monopoly movement in its core. State constitutions talk about monopolies interfering with freedom.
9:23 Speaker 05 Right, and so just as we won't tolerate kings, we shouldn't tolerate monopolies who tell us how we have to behave, what we see, what we consume. And so when we're talking about the free flow of information, which is vital to a political discourse, or we're talking about information that's used to target us as individuals, those go to the foundation of our democratic values and our fundamental freedom. And it's important that we're protecting competition, the competitive process, so that we have choice, so that we have freedom of choice. We can pick where we want to work, we can pick what services we want to use, we can pick where we want to buy things.
10:02 Kevin Roose? So in the old sort of model of antitrust, the test to apply to figure out whether something was a monopoly is, are consumers being harmed by this? This new antitrust framework you're talking about, which is sometimes called hipster antitrust, I don't know if you like that term or not, but that has been applied, what is the standard of harm? What makes something a monopoly if it's not rising prices for the consumer?
10:26 Speaker 05 Sure. I will first be clear, my kids do not consider me hipster, notwithstanding my cool shoes here. But the problem is that, how do you measure price, for example, in a multi-sided market with, where monetization doesn't take place in a completely transparent way? The antitrust laws exist because we believe competition, we believe choice, we believe variety is foundational to our freedom. And so when we focus on competition because we believe that the range of benefits that
11:08 Speaker 05 result from competition are good for society, then that's a much easier thing to measure, and that's a much easier thing to preserve, and that's our focus. That is actually more adherent to the founding of the antitrust laws going back to 1890. Some of these economic theories that are often put out there as to narrow the antitrust laws didn't exist when Congress wrote the law in 1890, and then again in 1914, and then updated again. But the concept of competition being foundational to our freedom did exist. And so that's what we're reeling is we're going back to the OG, we're not necessarily trying to change it.
11:50 Casey Newton? You know, over the past several years, Congress has considered any number of changes to antitrust law that I think might make your job easier. None of those have sort of made it across the finish line. I'm curious, do you feel like there are things that Congress could do that would let you sort of more fully realize this vision that you're laying out?
12:13 Speaker 05 Sure. So, for starters, I have to take a step back and say, well, what's my job? My job is to enforce the law as Congress writes it. And so whatever tools Congress gives us, we're going to use, and we're going to use as effectively as we can. And so right now, we have antitrust laws that have been on the books since 1890, 1914, and beyond. And we're using those tools as effectively as we can to realize the purpose of the antitrust laws. We've come out at the Department of Justice with full-throated support for congressional legislation that would allow us to enforce the law in even more meaningful ways. But ultimately, the decision is up to Congress, and my job is to keep my head down and focus on enforcing what we have while we have it.
12:57 Casey Newton? Do you want to tell us anything specific they could do, though?
13:00 Speaker 05 So Congress, in the last Congress, proposed legislation, for example, that would impact the ability of tech companies to favor themselves or exclude rivals. They talked a bit about interoperability and openness and making sure that as markets, tech markets develop, they become more...
13:20 Casey Newton? This is so like if Amazon is using data from its sellers to then inform what it's making under its Amazon Basics, House brand, they can't go compete with the sellers in the Amazon marketplace. That's one example.
13:32 Speaker 05 These are the exact kind of examples that the committees that investigated this up on Capitol Hill used and looked at to assess whether to revise the antitrust laws.
13:44 Casey Newton? The one that drives me crazy is Apple Music, right? Because Spotify has to pay Apple some really big percentage of all of its revenue through the App Store. Apple gets to run its own music service. It doesn't have to pay itself anything, right? And so to me, that seems so obviously anti-competitive. Is that the sort of thing, you know, I imagine you can't talk about that thing specifically, but like... Do you have the law behind you if you want to say, "Actually, yeah, that doesn't seem right. We want to file a case"? Or does Congress need to change something?
14:14 Speaker 05 Sure. So if we file a case, just to be very clear, we believe we have the law behind us, and we believe we have the facts to support the application of the law. Without commenting on any specific matter, I can say that a lot of the problems that we confront flow from conflicts of interest. And so if you own a platform and then you compete on that platform, now all of a sudden, you're kind of conflicted. And often, a lot of the problems that we see flow from conflicts of interest.
14:40 Casey Newton? Yeah. All right, Kevin, try to get him to talk about Google.
14:42 Kevin Roose? Yeah. So the DOJ has had two big investigations into Google. What can you tell us about those investigations and where they might be headed?
14:55 Speaker 05 Good. So we have two lawsuits that we filed. One that's set for trial this September that relates to Google's search business. And we are going to bring that forward before a court.
15:08 Kevin Roose? And what, just in very basic terms, like what is the theory of the case there? That Google, is Google's search business is anti-competitive because why?
15:16 Speaker 05 Among other things, Google went out and entered into agreements to distribute its search engine exclusively, for example, through a phone manufacturer or through a browser, and locked up the opportunity for other search providers to get the kind of scale they need in order to compete effectively.
15:36 Kevin Roose? Okay, so that's one case. The search monopoly case, what is the other case?
15:39 Speaker 05 So the other case is one that we filed quite recently in the Eastern District of Virginia that relates to Google's advertising technology business. And without going into the specifics of the case, one of the things that we've talked about publicly and that's in our filing is that advertising technology really more closely resembles a stock exchange than it does the old Mad Men-style business. So one of the things we cite in our complaint as an internal document for Google where it analogizes its business to, for example, Goldman Sachs or Citibank owning the New York Stock Exchange. And the reason this is interesting
16:19 Speaker 05 is because the advertising technology is sold through an electronic trading market. And there are more trades, I believe, in the ad tech ecosystem than in any other electronically traded exchange in the world.
16:35 Casey Newton? Now, that case seems much better to me than the other one. So why did you file it second?
16:40 Speaker 05 Well, I won't weigh in on that specifically. But we file our cases when we're ready. So we investigate and then we are ready to bring a case. We file it. And so we have a very careful process by which we investigate. We evaluate documents. We evaluate data. We consult experts. And then when we have the goods, we file a case.
17:06 Casey Newton? That is a really good case, though. Let's talk, though, about the fact that the search market arguably is getting more competitive now, maybe than it's been. We've been talking a lot recently about Bing. Bing said that they now have 100 million, I can't remember if it's daily or monthly, active users, a third of which are new, like since they launched this new chat feature. And one of the arguments against taking too strong a position on antitrust, and I mean, this is the sort of venture capitalists who I spend a lot of time talking to in Silicon Valley talking here. But what they will say is like, hey, look, if you just relax, eventually the invisible hand of the market will come along and introduce some new competition. And I will say, I used to be very skeptical of this idea. And I was somebody who was foaming at the mouth
17:51 Casey Newton? for y'all to come in and break up Facebook and WhatsApp and Instagram. And then three years went by and TikTok came along. And now the entire industry is panicked and they're racing to copy TikTok, right? So I'm just curious how you think about that idea that often if we just sort of procrastinate and do nothing, the market is able to kind of unwind some of these more anti-competitive situations.
18:13 Speaker 05 Is it? So...
18:25 Speaker 05 Let's think about it this way. What are some of the most important technological changes in the modern era?
18:33 Casey Newton? The invention of podcasting, number one.
18:36 Speaker 05 I had that at two, but we'll bump it up. So the breakup of AT&T is what led to the hardware revolution, including the patents and the IP and all of the and the growth and evolution of the infrastructure that led to the Internet. The case against IBM, the antitrust case preceded the software revolution, Windows operating system and the like. The case against Microsoft preceded the Internet revolution and the growth of companies like Google, Facebook, and Amazon. And so historically, if you look at key inflection points, in many instances, that is the most important time to protect competition. Because the incentive for a dominant firm
19:21 Speaker 05 to squash that inflection point in its nascenty is greater than it will ever be. And so one of the reasons, especially in technology, that we enforce the antitrust laws is to make sure that those new innovations have the ability to rise to the surface, have the ability to disrupt the world in ways that we cannot predict and don't want to predict. Because the beauty of innovation is the uncertainty. And that is what we're trying to protect.
19:47 Kevin Roose? Some of the challenges that have been brought to tech on antitrust grounds in the past several years have not worked out. Some of the cases that I'm thinking of, there was a case by the FTC that tried to block Meta from acquiring this VR company within that makes a fitness app called Supernatural. That was just, you know, sort of let through in court. Is it important to you that the DOJ actually win these cases? Or is just sending a message to tech companies that they can't do, you know, these kind of aggressive anti-competitive maneuvers, is that enough?
20:26 Speaker 05 I want to be clear about this. We bring our cases because we believe we are right on the law, and we believe we are right in the facts, and we bring our cases because we believe we're going to win. And one of the criticisms over the last 20 years is that the antitrust enforcement authorities haven't brought enough cases. They haven't given courts the ability to confront fact patterns that reflect a modern economy, that reflect ones and zeros versus poles and wires. And so in that sense, it's really important, if we're going to advance the law through the court system, that we give courts fact patterns, so that courts can learn, so that courts can weigh in on, for example, why targeting of individuals is different than consumers responding to price increases from one supermarket
21:10 Speaker 05 to another. These are very different markets that require different analysis, and unless and until we bring those to courts, give courts the opportunities to confront those fact patterns, we're not going to see the kinds of advancements in the law that we believe, and that I believe, are necessary to have a workable antitrust framework for a modern economy.
21:32 Casey Newton? You know, often when we talk about antitrust cases, we think about them solely in terms of, is a merger allowed to go through? But at the same time, in the past, y'all have been able to get creative with companies and just attach conditions to mergers that sort of make them, I think, more palatable. I'm thinking about, there was a case involving AOL Instant Messenger, where the government insisted that it be made interoperable with other messaging platforms. To me, that seemed like a really positive application of this kind of regulation. I understand you're entering every case to win, but I'm curious how you think about some of those other remedies, and whether they ought to be used more.
22:16 Speaker 05 That's hysterical. I worked on that case at the FTC as
22:19 Casey Newton? Oh, thanks.
22:20 Speaker 05 a second-year lawyer out of law school.
22:22 Casey Newton? Nice.
22:23 Speaker 05 So I think there are, there are, there's some opportunities and challenges. One is that, as an antitrust enforcer, I believe that our job should be to enforce the law. I don't view us as regulators. And so, the idea that we're going to oversee a company and make sure it, its code is structured in a certain way can be very difficult, particularly in the context of a merger. So merger law basically says, hey, If it's substantial less than competition, it may, rather, substantial less than competition, has the possibility you shouldn't merge and you should go compete instead. And that is a much better framework, in my view, to let the market figure it out through competition than trying to engineer certain outcomes
23:07 Speaker 05 through a regulation. And a criticism over the last 20 or 30 years has been that antitrust enforcement authorities have settled too many cases and those settlements have not worked effectively. For example, divesting a handful of assets as a carve-out in order to try to get market shares down to a certain level, rather than thinking about whether those divested assets will succeed in the context of being bought by a new company. And so I think it's really important that, you know, if we're going to accept a remedy, it really succeeds in solving the competitive problem.
23:42 Kevin Roose? So your office has filed lawsuits against Google. You've also taken on a big publishing case involving Penguin Random House and Simon & Schuster, reportedly looking into Apple and its App Store. What are some of the next things you think we might see from regulators, including DOJ, when it comes to tech antitrust?
24:04 Speaker 05 The fun part of my job is that I'm not allowed to talk about it.
24:08 Kevin Roose? Well, not normally, but here among just, it's just friends.
24:12 Speaker 05 Just friends, yeah, all right. Yeah, we let our filings do our talking. And so I will look forward. But I am trying to be and have tried to be transparent about the kinds of issues we care about. And, you know, I think about issues that have a really high impact on our livelihoods, on our lives. And technology, obviously, is, you know, core to so many things we do. I will point out, and I think this is relevant to discussion of both AI and specifically in technology more broadly, is that we use those terms as if it's a very narrow category.
24:52 Speaker 05 But these are just tools. So every market is in a sense a tech market, right? We're seeing technology issues in healthcare and in energy and agriculture. Where when we think about AI, for example, we think about AI as a tool that will be relevant, looking forward for, you know, all different types of applications. And so it's really important that we understand that, to the point where we're, we are hiring and have hired PhD data scientists, AI experts, and we're bringing in expertise to make sure that we have the ability to understand the technology we're seeing.
25:30 Kevin Roose? That's great. I want to, I want to ask one more question about that because something, something that you hear from people in government sometimes is that it's hard to attract the talent that you need to be able to go after these tech companies because some things like, like the ad tech market is sort of notoriously complicated and, you know, you can see these.
25:51 Speaker 05 Some have even said boring.
25:52 Kevin Roose? I would never accuse the ad tech market of being boring, but you'll, you'll hear people saying, oh, well, you know, we can't hire data scientists because they're, they're demanding such outrageous salaries, we can't convince them to work in government instead. So do you feel like your office has the technological expertise that you need to be able to bring these cases?
26:12 Speaker 05 Yes. So this is, I think, speaks volumes to the moment we're in right now. But we are attracting talent that is off the charts. People want to work on these problems. And so we are attracting our head of our expert analysis group helped start the AI Institute at Stanford Business School, and she's one of the top economists in the world. We are hiring former AI scientists, data scientists from, you know, some of the large tech companies. We are seeing, I spent the last three months touring law schools, and this is a little bit from the data side, and students are coming out in droves, not because they want
26:57 Speaker 05 to be the next Alex P. Keaton, but because they want to help solve these problems. Because they care about monopoly power. We have the benefit of having being on the right side of the issue. And people care. I think a lot of people I can see out here are part of this conversation today. Because they care about these issues, they see the impact. And so we've had the ability to extract, I'm sorry, extract, attract talent that
27:28 Speaker 05 that is, in my view, off the charts and has exceeded my wildest imagination. And we're building it from, not just traditionally as the antitrust division has been lawyers and economists and paralegals. But we're bringing in data scientists, we're bringing in AI experts, we're bringing in banking experts, we're bringing in healthcare experts. We're bringing in management consultants, people who can understand how businesses make decisions. In many respects, I think in order for us to be effective, we have to mirror the world in which that we, that we, the worlds that we investigate and where in the industries where we enforce the law.
28:09 Kevin Roose? We just have a few minutes left in this part of the conversation, but since you mentioned AI, I do want to ask you about AI because I think that's an area where a lot of regulators and politicians are looking right now. It's obviously a very fast moving area in Silicon Valley. And there's one theory that AI is going to be very good for competition because, you know, these things like ChatGPT are coming out of startups and they're, you know, Google is on red alert now and they're having to sort of hustle and be competitive for the first time in many years. As Casey mentioned, we're now talking about Bing, which seems to be a good sign for competition. So there's one theory that AI increases competition because it forces these entrenched, you know, tech giants to work a little harder.
28:51 Kevin Roose? And then there's another theory that actually it's going to decrease competition because you need so much compute and infrastructure and so many GPUs to run these AI models that basically only the giants can compete. And so you're going to end up with AI concentrating more power in the hands of companies like Google and Microsoft. So which of those arguments do you think is more correct?
29:13 Speaker 05 Yes.
29:14 Kevin Roose? Both.
29:15 Speaker 05 So, let me, let me give you some insight into how we're thinking about AI. So when I first started at the DOJ, it's a little over a year ago now, I recognize that AI is probably going to be one of the most significant issues that we have to tackle. So we, we started hiring up the experts. We have a thing we, we call Project Gretzky, which is...
29:36 Kevin Roose? Project what?
29:37 Speaker 05 Gretzky. As in Wayne Gretzky? Wayne Gretzky, yeah. Where, instead of skating, he has this famous saying about, "I don't skate to where the puck is, I skate to where the puck is going." And so we recognize that AI is going to transform how people do business. I think there are a couple of ways to look at this, or, or both ways to look at it. So one is AI as a technology, the other is AI as a tool. And so first, we want to make sure that we're preserving competition in the technology itself. And that could be everything from the hardware, the interfaces that run on top of the hardware, the APIs or interfaces relating to data sets and indices, and then everything that sits in between.
30:22 Speaker 05 And then there's the application of the, of the technology, and that application might be AI-driven technology. And AI-driven technology has the potential to offer lots of great benefits, but it also has huge feedback effects and scale effects. And so the bigger you get, the more data you have, the more data points you have to train your algorithms and the like, the more markets can tip.
30:47 Kevin Roose? The big get bigger.
30:48 Speaker 05 The big get bigger, but exponentially so, which makes it really hard to compete. So we are keeping an eye on all of that. There's an added dimension too, which is kind of fun from an antitrust perspective. One of the things that we enforce, both as civil and criminal law, is, are things like price fixing agreements or market allocation agreements.
31:10 Speaker 05 Back in the old days of the trusts, it would be the proverbial smoke-filled room with a bunch of fat dudes in three-piece suits, smoking cigars, fixing prices. And then it became a little bit more like folks on email, writing emails or texts back to each other about how to fix prices or allocate markets. As AI takes on a much greater role and as businesses conducted programmatically, now we're going to have a world where individuals aren't necessarily agreeing themselves to fix prices or allocate markets, but their AI is doing that.
31:51 Casey Newton? Yeah, Bing will just say, "Looks like you're trying to fix some prices. Would you like some help with that?"
31:55 Speaker 05 Or your two AI might start talking to each other and realize that it's efficient. And so, you know, these are the kinds of challenges that we're preparing for.
32:03 Casey Newton? We've always said we're going to keep the Hard Fork Podcast free because we don't want any trouble with you guys, okay? So it's just going to keep it at zero and hope that's okay.
32:11 Kevin Roose? Well, everyone give a big hand to Jonathan Kanter. Thank you for joining us.
32:25 Kevin Roose? When we come back, Jonathan Kanter, Casey and I play Bot or Not.
33:16 Speaker 06 Today's AI builders need infrastructure that scales with their ambitions. Iron is a vertically integrated AI cloud platform, delivering data centers, compute and software for AI training and inference, underpinned by its expansive portfolio of land and grid-connected power in renewable-rich regions globally. Iron's AI cloud gives direct access to Nvidia GPUs, scalable from hundreds to thousands. For builders, that means direct access to AI compute, rapid scaling and predictable pricing. Iron, AI cloud for the next big thing.
33:49 Speaker 04 Your sales team is at risk of missing quota. Don't panic, just ask Rippling AI. Built on your real-time people and business data, Rippling AI can pull metrics from Rippling and your CRM into a meeting-ready dashboard showing quota attainment, headcount trajectory and revenue to quota by region. In seconds, you'll see what's behind your quota risk and fix it before it's missed. Head to rippling.ai/hardfork to get the only AI built to give you full visibility and take complex actions across your whole business. That's R-I-P-P-L-I-N-G dot AI slash hard fork. Sign up today.
34:25 Casey Newton? We have one more treat and this is exciting. This is the audience participation portion of today's podcast.
34:32 Kevin Roose? Yes, so we have our first ever quiz game that is going to happen on this stage. And it's a little game that we developed called Bot or Not.
34:46 Kevin Roose? So in this game, we will present each other with a series of creations, media objects, if you will. And some of these I know the answers to and some of these Casey knows the answer to and I don't. And so we will quiz each other and with your help, we will try to decide whether these things were created by AI or humans.
35:09 Casey Newton? And together we can determine whether humanity still has any value left.
35:15 Kevin Roose? Yeah, this is sort of like, it's occurring to me that we basically designed a captcha.
35:22 Kevin Roose? Pick out all the red stop signs. Okay, so our first question, and you can help us out with this too. Are these real TikTok trends? Or were these created by an AI? I'm saying AI. Fake it hashtag, so we have three trends. Hashtag fake it till you make it. And I can give you some more details about these. This is a hashtag where you pretend to be someone else. Hashtag break the rules and the hashtag dare me challenge. Casey and Jonathan, do you believe that those are real viral TikTok trends?
36:06 Kevin Roose? Or did an AI come up with those?
36:08 Casey Newton? What I would say is fake until you make it. That actually sounds like a real trend. If it was just that on stage, this would be very difficult for me. I think the dare me challenge sounds exactly like what an AI would say if you said, give me a fake trend.
36:20 Speaker 05 Yeah, I'm gonna say AI. It just feels a little boring.
36:23 Kevin Roose? You think this is bot? What do you guys think? Bot or not? Bot. Bot? Getting a lot of bot. You are correct. This was Bing Chat's answer to the question. Imagine three viral TikTok challenges that might sweep the nation's youth, shocking and alarming parents and teachers. You gotta get a little creative with the prompts on these. I love that. Okay, here's our next question. We're gonna play an audio clip and you're gonna say, is this a recording of Elon Musk obtained through rigorous sourcing within Twitter? Or B, a synthetic voice sample that I created using 11 Labs new AI voice cloning tool? So let's play this clip.
37:02 Speaker 07 I can't believe these guys keep talking about me on their dumb podcast. I'm just trying to extend the light of consciousness and destroy the woke mind virus over here. What kind of a name is Hard Fork anyway? I'll show them a hard fork.
37:15 Kevin Roose? Okay, Casey and Jonathan, is this bot or not?
37:20 Kevin Roose? Casey?
37:22 Casey Newton? Oh god, I wish it were real. I wish it were real so bad. I'm gonna say no, but that was honestly, I think, the best synthetic voice we have played on the show so far.
37:33 Kevin Roose? Yeah, I agree.
37:34 Casey Newton? It was very good.
37:35 Kevin Roose? Well, no, it's not a synthetic voice.
37:36 Casey Newton? I'm sorry.
37:36 Kevin Roose? Maybe it is, maybe it isn't. What do you think, Jonathan?
37:38 Speaker 05 I think I want to be friends with that bot.
37:41 Kevin Roose? What do you guys think? Bot or not? Bot. Sadly, that was a bot. Although, if you have, if you're, any Twitter employees are in the audience today and you do want to leak us audio of Elon Musk talking about the Hard Fork Podcast, get in touch.
37:53 Casey Newton? I think, like, should we actually have, like, fake Elon Musk read the credits each week?
37:58 Kevin Roose? I think so, although he's very litigious, so I don't know if I want to go there.
38:01 Casey Newton? Good point.
38:02 Kevin Roose? All right.
38:03 Casey Newton? I know a good lawyer, though. All right.
38:09 Kevin Roose? Okay. Here is four quotes from tech leaders, three of which are real, and one of which was hallucinated by ChatGPT. So, Casey and Jonathan, I'll just read them out. Option one: "Don't follow your passion," by Marc Andreessen. Option two: "If people don't have more children, civilization is going to crumble," Elon Musk. Option three: "In the next two decades, we envision a world where we have permanent settlements on the moon and Mars, as well as other celestial bodies in our solar system," Jeff Bezos. Option four: "Spam will be a thing of the past in two years' time," Bill Gates. So, Casey, which of these do you think was generated by ChatGPT?
38:53 Casey Newton? I am going to guess Jeff Bezos, because I have trouble hearing him in my mind saying celestial bodies, and also, I don't think that even he thinks that that's going to happen within two decades, so that's my pick.
39:08 Kevin Roose? Jonathan?
39:10 Casey Newton? I'm choosing between two and three, but I'm gonna go with three.
39:14 Kevin Roose? Three Jeff Bezos, what about you guys, what do you think? Three. Three?
39:18 Casey Newton? We have some fours as well.
39:20 Kevin Roose? You guys are too good. This was, in fact, a hallucinated quote from Jeff Bezos by ChatGPT.
39:27 Kevin Roose? All right, we're doing pretty well. Next round. Here are two recipes. One of them is real, and one of them was generated by GPT-3. Option number one is roasted turkey with a soy ginger glaze. Option number two, shortbread cookies with olives and rosemary. Casey, which do you think was generated by GPT-3?
39:51 Casey Newton? Oh, well, I wanna say two because I hate olives.
39:57 Kevin Roose? The shortbread cookies with olives and rosemary?
40:00 Casey Newton? Yes, but I'm noting that the first recipe begins, "This roasted turkey recipe is inspired by the flavors of my childhood", which is like the most generic recipe intro ever that I think a chatbot actually would would use because it believes that that is how recipes are supposed to begin, is with 18 paragraphs of personal history.
40:23 Casey Newton? This is, this, I think this might be the hardest one.
40:27 Casey Newton? Let me phone a friend. How many of you guys think it's one?
40:32 Kevin Roose? One is, how many of you think the roasted turkey with the soy ginger glaze is the GPT-3 bot recipe?
40:37 Casey Newton? Maybe a third have hands to half hands, and then how many think it's two? It's like 50/50. It really is 50/50.
40:45 Kevin Roose? Even split. Jonathan, what do you think?
40:46 Casey Newton? Well, I still have faith in humanity, so I'm gonna say two. You're gonna say two. And I think I'm gonna say that one is fake.
40:53 Kevin Roose? Wow, okay. So Casey thinks that the roasted turkey with the soy ginger glaze was written by a robot. Jonathan thinks that shortbread cookies with olives and rosemary was generated by a robot. The correct answer is? The bot created roasted turkey with soy ginger glaze, the recipe from childhood. And the real recipe was shortbread cookies with olives and rosemary.
41:18 Casey Newton? And if you look under your chair, we've actually baked you, we've roasted you a turkey with soy ginger glaze. We were up all night roasting turkeys for you.
41:29 Kevin Roose? Thanks to our colleagues at New York Times Cooking, both for the real recipe and the GPT-3-generated one. For our last question, we have a special bonus round.
41:40 Casey Newton? Bonus round!
41:41 Kevin Roose? And even I do not know the answers to this one, although our producer has texted them to me, so I have them in my pocket. Bonus round. Here are six descriptions of reality game shows. Some of them are real, and some of them are fake. Were generated by GPT-3. So I'm gonna just read out the titles. Singles Inferno. This is a show about flirty singles searching for love on a deserted island. Then there's The Line. How long would you stand in line for $100,000? Contestants try to wait each other out to win the huge cash prize.
42:18 Casey Newton? Yeah, I've actually played that game trying to get brunch in San Francisco.
42:22 Kevin Roose? Ayo! Show number three, I Wanna Marry Harry. This is single American women compete for the heart of a Prince Harry look-alike, who they are led to believe is the real thing.
42:34 Casey Newton? Who goes on the show to win the heart of a Prince Harry look-alike? I hope that one's real.
42:42 Kevin Roose? Number four, Celebrity Toe Wrestling. This is celebrities compete in a toe wrestling tournament. The first celebrity to pin their opponent's toe wins. Show number five, Close Encounters of the Dating Kind. This is singles who believe they were abducted by aliens, go on dates with each other, trying to find a compatible partner who shares their unusual experience. And number six, Celebrity Swap, which is famous personalities switch lives for a week. Watch as a famous athlete takes on the role of a Broadway performer, or a celebrity chef tries their hand at being a professional wrestler. So Casey and Jonathan and our audience here at South by Southwest, which of these were created by a bot?
43:23 Casey Newton? All right, and do we assume that it's like three and three?
43:26 Kevin Roose? I don't think we can assume that.
43:28 Casey Newton? Okay. All right.
43:30 Kevin Roose? I don't actually know, but I don't, that's not in the description, so I don't think we can assume that.
43:34 Casey Newton? Here, we could work through these a few different ways. I would like to start by saying the one which I think is most obviously actually a show.
43:42 Kevin Roose? Okay, but hurry up because we only have two minutes.
43:43 Casey Newton? Okay, I think Celebrity Swap is definitely real.
43:46 Kevin Roose? Is definitely real?
43:47 Casey Newton? Yeah.
43:47 Kevin Roose? Okay.
43:48 Casey Newton? I think I Wanna Marry Harry. Oh, that is so plausibly real.
43:54 Kevin Roose? I feel like Quibi could have greenlit that one.
43:56 Casey Newton? Yeah, 100%. I'm going to say it's real just because I actually have trouble imagining like ChatGPT coming up with Prince Harry look-alike. Like, that seems like beyond it. So let's say that that one is real too.
44:09 Casey Newton? Singles Inferno is fake because it's too generic.
44:13 Casey Newton? And I think Close Encounters of the Dating Kind is generic too.
44:16 Kevin Roose? Yeah, that one looks a little, well, no, but that, I can see that on SyFy. I can see that on SyFy.
44:22 Casey Newton? I think that's fake. And I think Celebrity Toe Wrestling is fake.
44:26 Kevin Roose? I don't know, man.
44:28 Casey Newton? The first celebrity to pin their opponent's toe.
44:30 Kevin Roose? Jonathan, do you have any thoughts on these? All these sound kind of plausible to me.
44:33 Speaker 05 Yeah, they do. I'm going to say that Celebrity Toe Wrestling and The Line are bots.
44:37 Kevin Roose? Celebrity Toe Wrestling and The Line are bots. Okay, Casey?
44:41 Casey Newton? Celebrity Toe Wrestling, Singles Inferno, and Close Encounters are bots.
44:46 Kevin Roose? And I'm going to say that Singles Inferno and Celebrity Swap are bots. Okay, let me open up my text message here from our producer and see.
44:59 Casey Newton? These results certified by Ernst & Young.
45:02 Kevin Roose? Okay, so it looks like four of the six were by bots. The only real reality game shows on this list are Singles Inferno and Who Wants to Marry Harry?
45:18 Casey Newton? Marry Harry was real.
45:19 Kevin Roose? The Line, that was real. Celebrity Toe Wrestling, The Line, Close Encounters of the Dating Kind, and Celebrity Swap were all generated by AI. Let's give everyone a big round of applause for playing Bot or Not.
45:38 Casey Newton? Wonderful. Well, I think we learned some lessons and we got some food for thought.
45:44 Kevin Roose? Yes, and thank you to Jonathan Kanter for joining us. And thank you to all of you for joining us.
45:48 Casey Newton? Thank you so much.
45:49 Kevin Roose? Thank you so much.
45:50 Casey Newton? We'll see you next time.
45:52 Kevin Roose? Have a great South by Southwest.
45:53 Casey Newton? Have a great South by Southwest.
46:01 Kevin Roose? Hard Fork is produced by Davis Lamb and Rachel Cohn. We're edited by Jen Poyant. Today's show was engineered by Brad Fisher. Original music by Dan Powell and Marian Lozano. Special thanks to Jordan Cohen, Paula Schuman, Nell Galogly, Kate LaPresty, Jeffrey Miranda, and everyone who made it out to see us in Austin.
46:48 Speaker 00 You can't undo everything in life, but you can undo prediabetes. More than 2 in 5 adults in the US have prediabetes. Most don't know they have it. But with early diagnosis and action, you can delay or even prevent type 2 diabetes. Visit doihaveprediabetes.org to take the one-minute risk test and start undoing the things you can. Brought to you by the Ad Council and the Centers for Disease Control and Prevention.
Previous GPT-4 Is Here + The Group Chat Bank Run Next Bard Fork + How to Talk So Chatbots Will Listen Page updated October 26, 2026 · transcript and summary are machine-generated