AI & the future of media with The Atlantic CEO, Nicholas Thompson

Nick Thompson, CEO of The Atlantic, led one of the first major content licensing deals with OpenAI in 2024. In this conversation, he joins Azeem to unpack how AI is transforming media – and what that means for every business navigating the shifting economics of attention, trust, and discovery.

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Azeem Azhar?

  I see four horsemen challenging the business of journalism right now. There's the challenge of the sovereign creator.

• 0:07Speaker 01

  Power has moved from the center and from the brand to the individual. It's a really interesting trade off. Most writers, they're way better when edited. They may not think that's the case, but it is true.

• 0:17Azeem Azhar?

  The second horseman is the collapse of search traffic.

• 0:20Speaker 01

  About forty percent of our traffic comes from search. You know, we're seeing a significant decline, maybe a twenty percent decline. I would like a system that helps everybody in perpetuity, the way the old Google system worked. So if you want to crawl our site, you have to give us something. The third is weakening of trust and authority that media houses have had. This is something where AI can be really useful. Please read this story and identify like the six embedded assumptions that I don't even notice because I'm in some kind of a filter bubble that are going to trigger readers on the other side.

• 0:49Azeem Azhar?

  The fourth horseman is AI.

• 0:51Speaker 01

  That's my favorite. It's the most fun to talk about. It's less, I lose less sleep on that one. Is it a competitor? Is it a collaborator? Fact checking is hopeless. The copy editing stuff is pretty useless. They're amazing for certain kinds of editing suggestion.

• 1:10Azeem Azhar?

  Hello, everyone. I'm Azeem Azar, founder of Exponential View. And today we're going to look at the future of media, the future of journalism, with my friend, the CEO of Atlantic Media, Nick Thompson. Now, the last time Nick and I did anything on video, he put me in a weird exoskeleton and had me walk up some stairs in Geneva. I've been much kinder to you this time, Nick. Welcome to the show.

• 1:37Speaker 01

  I had forgotten that. It's true. Anybody fact checking, fact check accurate. I made Azeem wear an exoskeleton and I filmed it for the most interesting thing in tech.

• 1:45Azeem Azhar?

  Yeah, well, that was, that was a year ago and a lot has happened in the year. I see four horsemen challenging the business of journalism right now. I've put them into four buckets for, you know, for historical reasons. There's the, the challenge of the sovereign creator. Substack just raised a hundred million dollars at a billion dollar valuation. They have five million paid subscriptions and many journalists are leaving the traditional world to, to go out on their own. The second horseman is the collapse of search traffic and the rise of AI and ChatGPT as a source of traffic. It's now our third largest source of traffic. The third is a big secular shift.

• 2:29Azeem Azhar?

  The weakening of trust and authority that media houses have had, chronicled very well by Martin Guri a decade ago. Trust is becoming personal, not institutional. And the question is, what do people want from those commentators? And the fourth horseman is AI. Is it a competitor? Is it a collaborator? You were early to do a deal with the with OpenAI, but many laggards could be left behind. Which of these four are the biggest, most interesting challenges for the CEO of a media business today?

• 3:02Speaker 01

  They're all important. The most pressing is probably the collapse of search. It's happening very quickly. It's something we've known would happen for a long time. Right? We wrote our, we did our first study on how much search traffic would decline because of AI back in, you know, March of 24, so right after, um, right after the release of ChatGPT. I guess, no, March of 23, sorry. You know, so we've been trying to make our business more resilient and prepare it for a world where our top of the funnel traffic, meaning people are just coming in for stories, drops dramatically. And there's a lot we do to try to prevent that. So that's existential for many publications, less so for a publication like The Atlantic, but it's really important for us and we're deep in that.

• 3:43Speaker 01

  The first challenge, which is the rise of the individual creator and Substack. I don't think that as much as an AI-driven challenge, though actually that collapse of search, I think probably helps the rise of Substack in some ways. That was, when I started, it was the biggest concern we had. So I started my job as CEO of The Atlantic in 2021, and we were subsectioning to hire every Atlantic writer. It ebbs and flows and it's now, it's flowing. There's, there's a worry, you know, we've lost two very smart people, two wonderful writers to Substack recently, you know, and there are things that. You have to do as a traditional media publication to make your best, most ambitious, most capitalistic writers want to stay with you. And that is a challenge we're trying

• 4:25Speaker 01

  to figure out. Back in 2021, when it was a real risk, we started our own individual newsletter program, subscriber on the newsletter program. We brought people in from Substack. We sent sunset of that program, but it was actually pretty successful in certain ways. So that's that concern. The decline in trust, in some way, it's the most important because it undergirds all of this. It's also something where we don't have like our playbook to counter the decline in trust, to do the absolute best journalism we can in the absolute best ways, fact check everything as much as we possibly can.

• 4:57Azeem Azhar?

  In the face of that decline, of course, people don't care about the quality of the journalism in some sense.

• 5:02Speaker 01

  Right. And you know that with, you know, no matter what you write, Trump and Musk will go out and say it's false and they'll say it to people who will believe it. And you have to, you know, so there are very hard things to deal with. So, and then, oh, and then the fourth one, AI as collaborator. I mean, that's my favorite. It's the most fun to talk about. We can spend the most time there. And it's less, I lose less sleep on that one. I'm kind of, that one's, that one's like energizing and exciting. The others make me scream at night.

• 5:28Azeem Azhar?

  We're going to take you into your nightmares first. So, so thank you for that. Right. So let's start with the, the collapse of search, actually briefly, to, to help people understand how important search and discovery is for media entities. I think two different things have happened. One has been that social has stopped being a source of, of referral traffic from Facebook and then, then X with the changes after Elon took it over. And then the second is that Google is just, is starting to send less traffic to, to you commercially for publishers. That's important because that's what drives the eyeballs, that drives the advertising.

• 6:08Azeem Azhar?

  But can you give us a sense either from, from the Atlantic of what the impact on inbound traffic has been of the change in social and, and the change in search?

• 6:18Speaker 01

  Yeah. So, you know, about 40% of our traffic comes from, um, search. You know, for some publications, it's much higher. Maybe it's as high as 70%. You know, we're seeing a significant decline, maybe a 20% decline. And so multiply 20% by 40%, it's 8%. And so suddenly you're looking at a pretty significant drop in the number of people who come to your website. And that means the number of people who see your ads, the number of people who are going to subscribe, your brand awareness, right? So that's the decline in search. Decline in social varies more because there are many more contributors, right? And so Facebook actually has gone up since January for us, right? It went down for a long time and it's now gone up. Twitter, of course, is going down because Elon Musk is making it harder to click out the sites and he's turning Twitter.

• 7:03Speaker 01

  You know, Twitter used to be a link service and be a news reader. Now it's kind of TikTok. You go there and you're just sort of like sucked into videos from the Coldplay concert, right? But it used to be a place where you would go and you would find articles. So it was a great referral source. Now it's very small referral source. Countering that somewhat, you know, we'd seen our traffic rising on threads. We've seen our traffic riding on blue, rising on Blue Sky. We've seen our traffic going up some degree ebbs and flows on Reddit. So our social traffic is probably fairly similar to where it was a year ago. It's massively below where it was four years ago. Search is the, is the pressing concern right now.

• 7:35Azeem Azhar?

  And it seems to me that it, you know, some opportunity emerges from, from ChatGPT. As I said, within exponential view, we've seen it as our third highest external.

• 7:46Speaker 01

  Wait, ChatGPT is your third highest referral?

• 7:48Azeem Azhar?

  Yeah. Don't ask me how, I just look at the logs. The team tells me, they tell me it, it's just above, I think we go CertX, ChatGPT, then LinkedIn. I have no idea how this is happening, but it is, it's what the data shows.

• 8:04Speaker 01

  Wow, that's a, I'm going to learn about your answer engine optimization strategies. And it's like the 17th highest refer for us.

• 8:11Azeem Azhar?

  It might be to do with, you know, it might be to do with how Substack itself is, is playing in that space. It may be to do with. The nature of our audience and the nature of the ChatGPT audience relative to something like The Atlantic was 146 years old, I think. But we're definitely seeing it, and I know that the team is thinking about how do we turn that into a tailwind of traffic?

• 8:38Speaker 01

  Yeah, I mean, that's remarkable. Yeah, I'm just looking at our referral data right now. So right now on the site, you know, just at this very moment, there are 1,100 people who've come in from email. There are 370 who came in from Facebook, 310 from search, Google search, 149 from Google News, Google Discover, 73, Pocket, Bing, Twitter, you know, ChatGPT is even showing up.

• 9:00Azeem Azhar?

  I mean, it's interesting. So there we have this space where maybe there's a mode for new types of content optimization. But on the other hand, we've had this fascinating thing come out of Cloudflare. So Matthew Prince told us both in January, I think you were interviewing him, that in the old days, a search engine would index your site 20 times for every human it would send to you. And then recently, it was going up to a few hundred times. And then when you get to the LLMs, it's an order of magnitude worse, that the LLMs are constantly hitting your sites, imposing some marginal cost, but sending very few humans to them. And one of the things that Matthew has created within Cloudflare is, I think they call it the LLM jail.

• 9:45Azeem Azhar?

  So if you're an LLM crawler hitting a media site that's supported by Cloudflare, it will spin up all sorts of dead ends for you. And the hope is that they can build a business where the crawlers have to pay in order to access your site. I mean, do you think that that is a route that we go forward?

• 10:05Speaker 01

  I think it's brilliant. I mean, I talked to Matthew about it a bit. The issue for us, right? So we would like the crawlers either, we would like there to be an exchange, a fair exchange of value, right? They crawl our site, they collect information, they feed it back to users, they train their models, they make a lot of money off of it, we make nothing, right? In some cases, we've sued, you know, we're part of a consortium that is suing one of the AI companies. We've made deals with others, most prominently with OpenAI. You know, we would just like there to be some fair exchange of value. The best exchange of value would be like the one that media companies and Google have had for the last 20 years, where Google indexes you and then links people to you and then they go read it. And so who knows what the ultimate end state is, but in order for the media industry to be

• 10:47Speaker 01

  happy, there has to be some fair exchange of value, whatever it is. And so in order to get leverage in those negotiations, it is helpful if you can block the darn things from crawling your site. The way we've traditionally done it is, you know, we can put a little warning and robots TXT saying, don't crawl our site. You haven't licensed it. You're not giving us anything. And but they can get around it, right? So the advantage of CloudFlare is that they're really good at tracking these bots. That's what they do, right? They're great. And so they've built this handy little dashboard and I can go in there and I can say, okay, perplexity crawled your site this many times. Google crawled your site this many times. Apple bot crawled it. Claude bot crawled it. Weirdly, like the fourth highest crawlers from Huawei? I hadn't even heard of their bot. And so what we can do is we can just go in and say, no, block this one, block

• 11:32Speaker 01

  this one, block this one. And then we can go back, Apple, perplexity, whoever, and say, hey, we've blocked you. And in fact, we've blocked you with CloudFlare. So if you want to crawl our site, you have to give us something. You have to give us a link, right? You have to like prioritize, you have to do something, right? Pay it like some exchange of value, whether it's traffic, whether it's, you know, money. And what Matthew has done and what CloudFlare has done is given the media companies that were just getting tramped, right? The AI companies came in and they were like, you know what? Life's hard. Lawyers are expensive. Internet was meant to be free and also screw these guys. And they just went in and they took everything and they took it and they trained on it and they, you know, built their engine on it and there was no... They hid it all so that it would be hard for us to find, right?

• 12:16Speaker 01

  It was like a burglar comes in, takes your stuff and buries it underground, right?

• 12:19Azeem Azhar?

  Absolutely. I mean, very, very sophisticated networks of crawlers, right? To look invisible, to look accidental, serendipitous, rather than a big sign saying, "I'm coming in to hoover it all up."

• 12:30Speaker 01

  Right. And then not only that, like, we're not going to publish system cards that show what we were trained on, mainly so that the people whose stuff we took to train on can't, you know, don't have a, like, have a hard time coming after us. You know, they did that because they didn't have cash, negotiations are hard. But now they're big, they're worth billions of dollars. And so there's this original sin and, you know, they have to figure out how to deal with it and we have to figure out how to deal with it. And so what cauliflower has done is it's changed a little bit of the leverage. There are other ways the leverage could change. It could change through lawsuits, it could change through legislation. There are lots of ways the lever could change, but, you know, it has been unfair negotiations so far and now it's getting fairer.

• 13:09Azeem Azhar?

  Like you, I see this as a game, a game of leverage. There will be a settlement of some sort that will be, that will be arrived at. It'll be an economic exchange. It probably won't be as high as 30%. It probably won't be, you know, lower than, you know, 3% to 4%. It'll be in some range, it'll depend on leverage, it'll depend on politics. And, and we've seen that happen in other environments. It's been, you know, the cut of the music industry or the royalties that an author gets relative to the final sticker price of the book. I mean, there are, there are markers. Do you think it's better for it to be resolved through that kind of process? Or would you like there to be, you know, a group of economists to go out and figure out what a fair exchange ought to be and there to be a presidential commission that recommends some kind of output?

• 13:54Azeem Azhar?

  I mean, is it better for the market just to figure it out the way it is today?

• 13:56Speaker 01

  I think that the economists should go out and they should do research on what a fair exchange would be. Like, I would actually love to have, I can do a back of the envelope calculation. Like, what percent of the training data? Let's just take Anthropic as an example, because they've basically made no deals with anybody. What is the value of Anthropic? Okay, that Anthropic, $100 billion, right? $100 billion. What percent of that value comes from like, you know, the initial training of the model? I don't know, what do you say, $10 billion, right? So, you know, what percent of that comes from all of like, the media industry and creators? I don't know, 10th of that, right? A billion dollars, right? So, okay, and then what should be the sort of the split? I don't know, 50/50, right? So they're like Anthropic is like a $500 million obligation, right? Or some hundreds of millions of dollar obligation

• 14:39Speaker 01

  to all the companies that are trained on the authors whose books they use, the magazines, you know, the musicians whose music they listen to, all that stuff. So there should be some way of making deal like that. I would like a sophisticated economist to do a more sophisticated analysis than what you and I just did in two seconds. And that would be very helpful for helping to set the market. It would be very helpful for governments that are trying to figure out what the market should be. Ultimately, what I want, though, like, I don't want cash payouts. I mean, I, we would like to be paid. They're nice, right? Yeah. I would like a system that helps everybody in perpetuity, the way the old Google system worked, even the way the old Facebook system worked.

• 15:23Speaker 01

  And so you can see like perplexity, ChatGPT, you know, getting closer to that, like perplexity does have links, ChatGPT does have links, you're getting traffic from it. If you could actually get to a point, the best model by far is ProRata. And so what they do is they both prominently display links. So you ask it, you're like, hey, what is the biggest story in AI today? Right? And it's like, the biggest story in AI is XYZ. And then it's got a link to Azeem's Substack. It's got a link to an Atlantic article. And the links are prominent, great. And then two, it runs a reverse search to see or reverse AI analysis to see, well, to what degree was Azeem's content responsible for my answer? Oh, it was 6% responsible. Then it takes, you know, a cut of its revenue and it gives.

• 16:08Speaker 01

  You know, 6% of that cut to you, right?

• 16:10Azeem Azhar?

  Right. So that Pro Rata is Bill Gross's business. Bill is a long-term friend of mine and I think a couple of my readers of EV are actually part of the founding team of Pro Rata. It's a really, really interesting model. The question is whether they can sit in the choke point of the traffic, right? Because right now, the bulk of the traffic is coming through ChatGPT and then some that's going through Gemini and some that's going through DeepSeek, right? So if you're only seeing fractions, basis points of the traffic flow, it's quite difficult to make a difference in the market. But perhaps they construct a model that becomes an appealing licensing model for the big companies.

• 16:51Speaker 01

  Yeah, right now, they have the advantage of having the best idea and the best implementation and the fairer system. They have the disadvantage of not having quite as many users as ChatGPT.

• 17:01Azeem Azhar?

  But Bill Gross has been there before because, of course, it was Bill who came up with cost per click advertising with GoTo.com before Google had it. And GoTo was acquired, I think, by Yahoo. And then Google copied the model and it built that ecosystem that has powered the web for the last 25 years. So maybe the sequel, a little bit like Empire Strikes Back, will be at least as good as the original.

• 17:24Speaker 01

  Google wanted a advertised model. They don't actually want a paid creators model. So I don't know how we get there, but yes, that is exactly what we hope. For the same thing that happened then, maybe Bill, I think Bill kind of got screwed 20 years ago.

• 17:37Azeem Azhar?

  He did a little.

• 17:38Speaker 01

  Maybe a model where Bill doesn't get screwed, that would be ideal, but like a similar outcome.

• 17:42Azeem Azhar?

  A quick note, if you want to support us in bringing more of these conversations to the world, please consider subscribing to the show. Okay, so I want us to get on to the Substack question. I've been on Substack for eight or nine years now. I forget, we started before they existed. We've been there a long time. I think we were the second or third paid Substack.

• 18:04Speaker 01

  Eight or nine years?

• 18:05Azeem Azhar?

  Yeah, yeah, we just crank it out. And we've got to know the team really, really well and see how they mature. That they are a really, really great founding team, very respectful. They make mistakes, they learn from their mistakes. They have had an inflection point, you know, they had about a million paying subscribers across the board. A year or so ago, they're up at five million now, and the model is that they take 10% of the revenue from each subscription. One of the things that struck me, and this does connect in a sense to our questions of trust and business models, is that you know, Substack-funded pundits like me are being paid to be believed, right?

• 18:50Azeem Azhar?

  It's about my reputation. It is moving from a world of selling attention to advertisers to selling conviction to subscribers. And that is a different way, I think, of thinking about the relationship between, you know, an expert and the people in their network and in their in their community. And what always struck me, distant past, I was a journalist, is that there was a very there were different types of journalists, and there were those who, like Derek Thompson, and like, you know, Anne Applebaum, or Tom Nichols, who I love, these are all, you know, Atlantic or former Atlantic writers, who had

• 19:31Azeem Azhar?

  that angle, that axis, around which they would have conviction that would be of interest to an audience. And there were a lot of journalists who were very, very important in the ecosystem, but really they were almost mechanistic in the work that they were doing. You must look at that across your talent and see what's happened with Substack. I mean, what is the kernel of how you think about it today differently to how you thought about it three years ago?

• 19:58Speaker 01

  Well, I don't think there's any doubt that power has moved, you know, it's true in media, it's true in all kinds of, it's even true in the NBA, right? Where you follow teams less than players, like the All-Star game, it's Team LeBron versus Team Giannis, it's not east versus west or whatever. You know, you see it in all kinds of industries where power has moved from the center and from the brand to the individual. And so You know, we clearly see that at The Atlantic where, you know, many people come in because of our brand, right? With this great brand, right? You're founded in 1857 by Ralph Waldo Emerson, like, pretty awesome. Like, most media brands, you kind of like run away from your founders a little bit because maybe they're like a little bit racist or... But ours were amazing, right? Like, Frederick Douglass is there, Harry Beecher Stowe is there. So, we have this incredible brand. We've been around forever. Some people love that.

• 20:42Speaker 01

  But yeah, a lot of people come in for the individual writers. And so, our challenge at The Atlantic is, how do you make the individual writers stay? How do you give them incentives? How do you make sure that you're giving them the editing support, the social support, the cultural support? That it's more that the incentives for them are high? And then we think about how do you compete with Substack? And in some ways, like, wait, imagine, imagine if you could have a Substack where you got Anne Applebaum, and Tom Nichols, like, and David Frum, right? Like, and Jeff Goldberg? And it was $80 for just that subset? Well, it's like The Atlantic, you got 100 of them. And, you know, so from a reader perspective, subscribing to The Atlantic versus subscribing to a whole bundle of Substacks is really appealing. What Substack does, though, is you have a direct relationship in a way that you deal with The Atlantic, right?

• 21:26Speaker 01

  Like, you can subscribe to The Atlantic and you get all of Anne's articles, but you don't feel as connected to Anne as you would if Anne had a Substack, right? And I think, you know, Derek, who has moved over to Substack, right? Part of the issue is, as I'm sure you know, the best Substackers make hundreds of millions of dollars, like you, Azeem, right? And, you know, there's a chance to, like, make more money than you can in a traditional media organization if you are a hugely successful individual. So, anyway, our challenge, I like the Substacks founders. I think it's a great product. I think they've done a really good job. I think it's kind of amazing that they got there and beat Medium despite Medium's head start in so many ways. Like, that is a remarkable business story. And then, I just got to make sure that the incentives

• 22:12Speaker 01

  never allow another Atlantic reader, or writer, to go over there.

• 22:15Azeem Azhar?

  Well, you know, there are a few other things that one gets. There is a sense of freedom. There is a sense of...

• 22:25Azeem Azhar?

  Just being able to say what you want to say, when you want to say it, being able to go really deep if you want to, being able to be very personally disclose disclose the way you're thinking and also just being quite short and here's a paragraph of of something. And I think that that degree of freedom is is really quite revealing and it and it's something that I felt a great deal. And I think there's just one other thing, which is that I obviously every writer benefits from editing. And I will say, of course, you know, exponential views and my essays and my books are no different from that. They've benefited from editing.

• 23:05Azeem Azhar?

  But there is something about about the freedom and being able to bring three decades of expertise out on the page for people that I really struggle when I've written op-eds for the New York Times or the Financial Times or or other outlets.

• 23:19Speaker 01

  Yeah, it's a really interesting trade off, right? Like, some writers probably are better. Some readers probably prefer writers who aren't edited, right? Because you get them like unfiltered, you get it right at the moment, there's no time delay. Most writers, they're way better when edited. They may not think that's the case, but it is true, right? Like, you know, I spent many years at The New Yorker. Take all those writers, put those first drafts on the internet, right? Show them to readers. The writers might be happy, the readers won't be happy. And there's a real value that comes from that. There are individuals where it makes more sense to them for them to go raw, unfiltered, direct. And then there are many where it makes more sense, like, I love the fact checking, I love the editing for my prose. But I think you're right. You know, there's some people where they prefer and it's better

• 24:02Speaker 01

  for their readers. And for those folks, Substack is right. And then there are a lot of folks where having the support of an institution is is more valuable. And so I think this world will shake out into a place where there are amazing Substacks, obviously, right? And they're continuing, it continues to grow. And the fundraising round was really impressive because You know, they've been struggling and it looked kind of like they were going to head for a down round. And when you head for a down round, you have problems. They clearly picked it up. Like whatever decisions they've made, the way they built in the social integrations, like props to Hamish, props to that team. Like that is like, you know, they've done some very good work there. So they figure that out. But I also think that this world shakes out with Substack's success and also the success of a bunch of publications that

• 24:46Speaker 01

  aggregate writers that add editors and that do something different.

• 24:49Azeem Azhar?

  You know, Substack in its own microcosm is a little bit like that. So the two most successful Substackers are Heather Cox Richardson, who is an American academic, a historian, who had a big Facebook following and then went on to Substack and has an enormous subscription base. I mean, it is institutional grade scale. And on the other hand, you have Barry Weiss, who left the New York Times, built something called the Free Press, which is an aggregator built on Substack. And it looks like a webzine, to use an old word. And it is exactly that. You know, you get your single subscription, you get podcasts, you get essays, guest essays,

• 25:32Azeem Azhar?

  on-staff essays. And somehow she's parlayed the low entry cost of being able to set up a media business on Substack into something that looks more traditional in terms of its organizational structure and its tempo and its cadence. And Substack has been able to to support both of those. And in some sense, that means, you know, there's nothing new under the sun in a way. We're just seeing a market segment in different ways.

• 26:00Speaker 01

  And you're seeing a really interesting thing, which is true, I think, across all new media forums, which is particularly in the case of Barry, where because traditional media, you know, very much so in this country, has been quite liberal, right? And the people who go into it come from a Relatively liberal backgrounds and it's been relatively tricky for a conservative or someone not that conservative. She had a hard time culturally at the New York Times because of the, you know, sort of the accepted ideologies within the paper, right? Stories she's told many times. What you've seen really interestingly is that smart conservative journalists have gone out and like, why do conservative journalists dominate on YouTube? Why do they dominate on Substack? Why do they dominate on all these new media forums? Partly because they're innovative, creative, but also partly because they haven't been

• 26:45Speaker 01

  able, like, they're not able to make their careers at the New York Times. And so that's a really interesting lesson for a place like The Atlantic where you have to, like, we have a much broader spectrum of writers and staffers than certainly any other media place that I worked, any other place in New York media. But if you don't have a really broad spectrum for you, welcome. Well, then they're going to leave and then they're going to start competitors in new forums and you're going to be in trouble, like Barry on Substack.

• 27:08Azeem Azhar?

  Right. And, and you know, I think that conservative commentators are incredibly strong on YouTube relative to progressive or liberal, liberal commentators. And I think, you know, the, within Substack, it's hard to know because algorithms hide, you know, the full spectrum of, of what's going on. One of the real powers of Substack has been that it drives subscribers to you. So I forget the, the number, but it's certainly a majority, or if not a majority, approaching a majority of all of our new free and paid subscribers have already been on Substack and are already subscribed elsewhere. And I think that that's a really powerful engine that, and a discovery engine that, that exists in, within Substack.

• 27:52Azeem Azhar?

  And people break out, they'll write a great essay, they'll have a thousand free subscribers. That essay will get 800 likes and reshares and they'll, they'll see themselves hit that 10,000 number very quickly. So there's something about the network effect within Substack, the, the discovery and its ability to drive traffic, which is something that media organizations used to provide writers, you know? So why would you write for The Guardian or The Times? Well, you'd get a salary, but you'd also get read, or at least you'd, you'd get distributed. And I think it's quite impressive the way they have figured out how to get eyeballs in a, in a way that doesn't feel like you're being algorithmically filtered to other writers, other essays.

• 28:34Speaker 01

  Yeah. I mean, and that is one of the things that Substack has done most effectively, and I think they have benefited. By the bifurcation of social media, right? As social media has collapsed, it has allowed the sort of the social network that Substack built underneath it, which seemed, it seemed totally barren, right? Like the sort of the feed and the recommendation stuff a couple of years ago. Now it's thriving, right? And that's partly due to the bifurcation of social media. It's due to like, you know, the way Elon has, you know, changed Twitter. So that is all true. I will add though, that traditional publications, like if you write for The Atlantic, we will take your essay and we will, you know, put it in, I don't know, a million, four subscribers, 600,000 of them get to print you. We will like use the United States Postal Service to deliver people. Like there are still ways that you can get, you know, we do have social feeds, we do have websites.

• 29:17Speaker 01

  Like there's still many, many advantages of traditional media. But yes, the social underpinning, again, back to the Medium example, I remember when Medium did that, like, and they started creating like an internal recirculation mechanism. I thought it was brilliant. I don't know why it didn't work there and then why it's ended up working at Substack.

• 29:34Azeem Azhar?

  There will be a business school case study on, you know, third time successful founders getting into this business and, you know, competing against, you know, first time founders or which some of the Substack people were. But, but I think there's something else that is, that is going on, which is connected to the, the shift from institutional trust to personal connectedness. And I was talking to Jasmine Sun, who was an early Substack product manager, and she's now gone off on her own, absolutely brilliant, Stanford English undergraduate. So, you know, character you would recognize yourself as well. And she said to me that something that was really prescient and insightful, rather insightful. I said, look, I've invested in lots of companies, lots of startups, about 40 of them.

• 30:18Azeem Azhar?

  I will disclose when I've invested, but I think it's really high signal for my readers to know that I have enough conviction in this theme, in this team, in these founders, in the way they're going, that I'm putting, risking my family's capital in something like this. And she said, look, there's a distinction in the East Coast. In the East Coast, that's a conflict of interest. And on the West Coast, we start to see it as skin in the game, right? That it really, really doubles down on the authenticity of the message. Because you have taken some risk. And I think there is something about that shift in how people can relate to institutions with great mastheads

• 30:59Azeem Azhar?

  compared to humans with, you know, eyes and eyebrows and sweat and nerves that actually feels like it's quite a difficult strategic challenge for businesses that have been, you know, built under a masthead.

• 31:15Speaker 01

  I mean, that's a really, that's a really complicated one. And your distinction between East Coast and West Coast is an important one. You know, I'm still, I'm still more on the East Coast side of that debate. I mean, I do think disclosure is important. I do think it's really important for, you know, we at The Atlantic are trying to figure out how we can make partnerships, equity partnerships, like, and disclose them in proper ways. There are ways that you can go off the rails. And I think that we're, you know, some of the West Coast model, you can see that, right? So the, my favorite example is I had a reporter, maybe my previous job, I was the editor of Wired. I had a reporter who was recruited by a venture capital firm. And they said, hey, come work for us. And she said, what will I be doing? And they said, well, mostly you'll be freelancing and you'll just be yourself. But maybe like 30% of your stories will be positive stories of our portfolio companies

• 32:00Speaker 01

  and you won't disclose it, right? You'll just go out and you'll write stories about that. And that, that's corrupt, right? What gets to the mindset where that can happen, right? And there's a slippery slope. Like, there's a moment where it's like, you know, Nick or Azeem invests in an AI startup, mentions that in an essay, right? Like Pro Rata, right? We have a partnership with Pro Rata, right? Like, The Atlantic is like a partner of Pro Rata. I praise it to hell and I mention the partnership. That's like totally fine, right? In the game, you know, there do have to be ethical lines. And so my hope is that in the future, I sometimes get a sense that the kind of the West Coast idea takes pleasure in blowing through some of the rules of the East Coast that are there for a reason. And what you want is you want

• 32:46Speaker 01

  Innovation, ideas, but also to like, to recognize that there's a reason why some of those rules about conflict of interest are there.

• 32:56Azeem Azhar?

  I would agree that the rules really, really matter in certain classes of, of writing and analysis. And what's on Substack is not just journalism. I don't think I do any journalism. I do essays, I do think pieces, I do trends and forecasts. And there are people who do real journalism, like proper investigative journalism. There's a, there's a very, very well-regarded British journalist called Jim Waterson, who left his storied newspaper to set up a Substack just to write about London. And he's taken the tools of being a great investigative journalist to cover stories that were simply not being covered because actually before Substack, before AI, local media was, was struggling because

• 33:41Azeem Azhar?

  classified had gone on, you know, on, on, onto eBay. So I think, I think those people do exist. I'm trying to square Martin Guri's prescient essay, this, the notion of the revolt of the publics in a world where there's so much information around that trust in institutions will start to decline. And I think the, you know, the Edelman Trust Barometer, the Reuters Institute for Journalism, their trust numbers for media are declining, but they are also stabilizing in some sense. Yeah, they're lower than they were 10, 15 years ago.

• 34:16Speaker 01

  Yeah, but you can't find too much further. I mean, like.

• 34:18Azeem Azhar?

  Right. But, but you know, is that, or how do you, how do you think about closing that challenge between increasing trust for better or for worse in talking heads and more skepticism about anyone with a masthead?

• 34:35Speaker 01

  I view it as. A real challenge and that our, our position has to be like, we have to fight to win and earn your trust every day. And then to also recognize that no matter how well we do, and no matter how hard we fight, and no matter how hard we try to be as honest as possible at everything, and as truthful as possible about everything, that we will still lose in certain ways because of the forces that Gury, you know, identified. The other thing that I think is important is that a lot of reason trust declined is because of the sort of the shift of the media ecosystem into filter bubbles, right? It happened because of sorting algorithms and for all kinds of reasons. And that doesn't just mean that you blame it. It means you also have to be self-aware.

• 35:19Speaker 01

  And so, back when I was a journalist, whenever I wrote a story, right, I would either read it to someone who disagreed with the principles. Like, I'm going to read this story out loud to you. I know you disagree with it, but I want you just to challenge every assumption in it, right? If it was a political story, I'd read it to a friend who's a Republican, right?

• 35:36Azeem Azhar?

  Back in the days when Democrats had friends who were Republicans and Republicans had friends who were Democrats, which I think is, is not the world the US lives in today.

• 35:43Speaker 01

  Unfortunately, that's, that's unfortunately true. Though, and this is actually something where, you know, again, I'm not a journalist anymore. I just, I just do the money, as my kids say. But this is something where AI can be really useful, right? So if you take a piece and you say, hey, I actually have done this with my book, right? Which my book is about running. It comes out in October. But like, imagine that you are like, totally skeptical of this person. Imagine that you like, want to find every error you can. Please read this essay and write kind of the like, most obnoxious but intelligent thing you can, right? And then that gives you a guide to like, oh, wait, let me harden this argument, or let me rephrase it. And it's super useful in politics, right? Please read this story and identify like, the

• 36:27Speaker 01

  six embedded assumptions that I don't even notice, because I'm in some kind of a filter bubble, that are going to trigger readers on the other side, right? And if you do that exercise, you realize that like, oh, that little throwaway comment about like Brendan Carr, right? Like, just to pick a random example, is going to just piss off all these people, and you didn't even notice it. So take out that stupid little line about Brendan Carr, right? Like, and that is something that AI can be super helpful with. So this came to mind because, you know, I used to, one of my old projects, I was trying to create a alternative social media ecosystem called Narwhal. It was like this fun experiment. We ended up calling it. But one of the premises was Like, we were trying to build an AI model that would challenge all of your assumptions

• 37:12Speaker 01

  and all of your political assumptions, so that before you would post, it would say, hey, by the way, here's a counter argument, would kind of like, sort of run the Reddit change my view mechanism through everything you say. We thought it would be a great way to teach people and ended up, we didn't get it off the ground, but whatever. But I do think it's a really useful exercise for journalists and writers and editors.

• 37:30Azeem Azhar?

  I love that example of having a critic, a helpful critic, to challenge perspectives on a piece. And I think that's, I'm mentally logging that to think, you know, we don't write much around politics, but lots of issues today have those, those political undertones. I found it extremely difficult to find that, that balance, because I think, as you know, you've, you've read a few of my, my essays, I really endeavor to, of course, I'm, I'm involved and I'm, I'm, I have a position, but I endeavor to have a bit of dispassionate perspective on, you know, the, the left and the right, the up and the down. And one of

• 38:12Azeem Azhar?

  the things that I find really tricky is that if I am critical of something that Elon Musk might have done, the inbox fills up. And if I then write about something that I think he's done that's really impressive, SpaceX, the inbox fills up and it's not the same people and they're all cross. And so there are these hot buttons where one can't at the moment, find a way of getting people to sit around the table and say, listen, we can accept that this, you know, a person does something impressive here and they do something difficult here. And that's all within the capability of a single person, personality, character. And I just wonder, you know, are there historical precedents for us going through these, these moments? And what are the things that we can learn

• 38:54Azeem Azhar?

  that help us come back together, bring it back to, you know, to a sense of, of common understanding while still being able to disagree quite civilly?

• 39:04Speaker 01

  I don't know. I mean, yes, there have been like, you know, I don't know, we used to have, newspapers used to be, you know, much more toxic to each other. People used to like punch each other out in the United States Senate. So like, there have been like really rough moments in, in American history. Part of what

• 39:19Speaker 01

  You know, we had a period of more civil understanding during the Cold War, but that's because we had a massive external enemy. We were worried the Soviets were going to nuke us. And so,

• 39:28Speaker 01

  you know, you had people who, you know, I wrote about George Cannon and Polnitz, that people who disagree with each other profoundly and like genuinely believe that the other's ideas will lead to nuclear obliteration, but can maintain a friendship. And part of what allows you to have that communication is the external enemy, right? And so if some people actually say this will happen with AI, AI, right? Like, what could possibly bridge not just American, American Republicans and American Democrats, but like, what could bridge like nine states and China? Like, well, maybe if there's like an external like catastrophe coming from, from AI. But I also think there are mechanisms, like this is something, when I was in college, I started a newspaper with the head of the conservative paper, right? And like, you know, I just think that people, you can work at it and you can do it. Back

• 40:10Speaker 01

  to your Musk idea, when you were saying that, I was wondering, like, I wonder if the next time Azeem writes about Musk, you should like, let's say you're, you're writing about SpaceX, like, put the essay, ask the AI bot, imagine that you're somebody who hates Elon Musk, right? Go through all of my assumptions and see whether there's anything I've made that will like trigger them. And then do the reverse when you're writing about like ketamine, right? And, um, or when you're writing a regular essay, like-

• 40:35Azeem Azhar?

  I don't write about ketamine that much. I mean, we've written about psilocybin, we've written about GLP-1s, but I, I mean, I love, I love this idea. We do, I mean, we do use AI tools in that way. I mean, we use AI tools to, to challenge perspectives, to find weaknesses in arguments.

• 40:50Speaker 01

  By the way, I love your, your posts on how to use, your posts on how to use AI tools are always awesome. They're great. They're some of my favorite exponential view posts.

• 40:57Azeem Azhar?

  Well, on Saturday, we've got a really interesting one that I think is, plays into to the limits of AI. So, so on, earlier this week, we, I wrote something about Donald Trump and a Senator McCormick putting something together to activate tens of billions of dollars in an industrial policy to generate more electricity. US turning to industrial policy. And we described McCormick as a Senate hopeful. He is a sitting senator since January and human fact checkers are there and there are LLM fact checkers. And the LLM fact checker didn't spot it because it thought that the big piece was, you know, the $80 billion, which is, which is true in some sense.

• 41:41Azeem Azhar?

  One of our readers pointed this out, as we started to diagnose what went wrong, I put more and more sophisticated prompts into different LLMs, Grok, Perplexity, ChatGPT's O3, and after six iterations, it was still not spotting the mistake, even though I was saying, check every job title. Assume we have described every person wrongly, check every single one. And it was a real, it was quite an interesting lesson to us actually about the limits of some of these, these tools, which we use really, really extensively. And we've, you know, we've learned a little bit from, from that. And but, you know, my, my sense is that, that it's quite hard to use AI well

• 42:25Azeem Azhar?

  in research, in writing support, in steel manning and challenging ideas. But these tools are being used pretty extensively by writers of all types right now. Is that your sense as well?

• 42:36Speaker 01

  Yeah, so I've, I've tried both for sort of professional reasons. They want to know where this is going and also because I've had this book. And my experience is that they're amazing for certain kinds of editing suggestions, right? Like, you know, this thousand word section feels a little bit slow, right? Please read the entire manuscript and help me understand like, what are 150 words I could cut out to move the reader along, right? Like, here is a paragraph that I don't think fits here chronologically, but there's this one quote that's really important. Please identify another place in the book where this could go, right? Like, that kind of stuff, it's amazing. Fact checking is hopeless, right? In part, like this date problem, right? Because it doesn't understand, it has a hard time parsing like, yeah, McCormick was, there are more stories that were written about McCormick

• 43:20Speaker 01

  when he was a Senate candidate than he was a senator. He's kind of disappeared, but we cared a ton about his race. So like, there's a bias in the training sample. Like, it'll probably be like 10 years before AI knows he's actually in the Senate, right? Because that race was so, right? So the fact checking stuff is useless. The copy editing stuff is pretty useless, right? But the thought partner, you know what my favorite thing I did with AI? I mean, I'm sure you've done this, it's an old hat to you, but I went through and I had interviewed this woman for my book like seven times over the last six years. And I input the, maybe there's 3,000 words on her in the book. I input those 3,000 words and then I input all the transcripts of all the interviews. And I said, you know, please make sure everything squares with all those interviews. Please identify any quotes that I haven't used that are thematically similar you can put in.

• 44:05Speaker 01

  Please identify any material that I've missed, right? It's just, it's so hard to like, like pull those old transcripts and go through them. And it's amazing, right? And AI does that instantaneously, right? It's like a brilliant researcher that does quickly. So, you know, I agree entirely with your example. And I also agree entirely that it's like, one of the themes of exponential view is like, use these tools constantly and learn, both because it helps you and also it helps you understand it.

• 44:28Azeem Azhar?

  I think that your example is, also speaks to the, the general barbelling of talent and the general barbelling of, of who is succeeding in, in media. So you have to be quite sophisticated to use the tools quite well. The very, very best sub stacks are doing a thousand times better than number 500. The media companies that are strong in this market, like the Financial Times or perhaps the Wall Street Journal have got, or Bloomberg have got very, very, very distinct, narrow focuses from which they are building. And those who, who sit in the middle, traditional media, right?

• 45:11Azeem Azhar?

  From the Latin, in the, the middle are struggling because the seas are parting in some way. And I see this pattern recurring time and time again. And it's been a recurrent pattern from the days of the internet, you know, between the very big windows and small niches that are low cost and everyone in the middle is struggling to, to, to make a dime. And, and if you're, you know, as you run the Atlantic, do you, if you see that, which you may not do, but if you see that, how do you think about? Where your position needs to be, do you try to go big Bloomberg style? Or do you have to find a really, really clinical niche that you can own and live off?

• 45:53Speaker 01

  I mean, like, yeah, I'm terrified of being in the middle of the barbell. Like, I gotta get to the right, I gotta get to the big side, right? The Atlantic needs to, it's what we've been trying to do, right? We went from a couple hundred, you know, when I started, we were at 850,000, 800,000 subscribers. Now we're at 1.4, like, we gotta get to two, right? Like, we've got to be bigger, we've got to be successful, we've got to consolidate our wins. You know, because you're in the middle, you're toast. And so, totally, that's where I'm going for it.

• 46:15Azeem Azhar?

  But it's about the message, it's about the aboutness of the publication. So I have a strong sense of the aboutness of the Financial Times, a strong sense of the aboutness of the Wall Street Journal. I have a strong sense of the aboutness of some of the best substacks, like cynicism by Bill Bishop, or The Free Press, or Dylan Patel, who's moved off a substack with his semiconductor analysis. A lot of media companies had this odd aboutness, you know, we are sort of about the north bit of Chicago.

• 46:48Azeem Azhar?

  What is the core aboutness of the Atlantic in this world of declining trust and sovereign creators and content overload?

• 46:55Speaker 01

  That's a super good question, right? So we do have a disadvantage that we've always been a general interest magazine, right? We were founded to be a general interest magazine, you know, politic, literature and the arts, right? So we are supposed to be about a lot of things. And that is not like general interest publications tied, Newsweek, a lot of them are kaput, right? And so we're not going to give that up, we're not going to specialize, we're not going to become just a political magazine or just a science and tech publication. We are going to stay true to our nature, like, we have to be what our founders said we should be, right? And we will be that, you know, as long as we can be that. So that said, you then have to figure out ways that you can really distinguish yourself. So we distinguish ourselves, we hope. By A, being a publication for the very best writers in the world, right? And we have a real diversity of writers. We let them take their own voice.

• 47:39Speaker 01

  We let them write their own thing, right? We also are a place for incredible investigative, long stories, right? And so our hope is that we are known. What we want, and what I want, is I want someone to go to atlantic.com or to pick up The Atlantic and to know that everything they read will be interesting and smart, that their assumptions will be challenged, that the writing will be beautiful, that they'll be taken to places in their mind they couldn't imagine. So we want this like constant process of discovery of awesome journalism. So we are trying to be the greatest writers collective in the world, right? That's the goal. And that, in some ways, is perpendicular to the goal of like being totally distinct. But in a way, it's not because it is like trying to be something that nothing else is. So that's the goal.

• 48:23Azeem Azhar?

  In a world where there's AI and automated content generation, having something that brings humans' abilities to imagine, extrapolate beyond what has been written previously, and distilling it and supporting that feels like it's distinctive. You're in a world of manufactured cheese. Artisanal cheese has all the value that we could ever, ever hope for. I know that we're at the top of the hour and we ran through the four horsemen. I feel like you have a lot on your plate to do that. But on the last note with your book, the book on running, you are a record holder in ultra-long distance running.

• 49:06Azeem Azhar?

  What is the one thing that you learned about yourself in writing this book?

• 49:10Speaker 01

  In some ways, it's the initial realization which triggered the book. It's gonna take a minute to explain, but I think it's pretty interesting. So I, when I started running, tried to break a three-hour marathon. And when I was 30, I ran a 243 marathon, which was like very cool, right? Right afterwards, like a week afterwards, I was diagnosed with thyroid cancer. And you can, you know, surgeries, multiple surgeries later, hard time. Two years later, I come back and I run the New York City Marathon again. And I run it in exactly the same time, 243. Okay. Then the next 10 years, from when I'm basically 33 to 43, every marathon is almost exactly the same. Like I literally go 243, 242, 243, 244, 245, 246. I have a like nickname, Mr. 243.

• 49:56Speaker 01

  Which is a little annoying, but still kind of, it's cool. Like, it's a good time, right? If there are runners out there, like, that's a good time. And then when I'm in my mid 40s, I start training differently. And I run a 229, right? In one year, I go down to 229. So suddenly, I'm like, now I'm, whatever I was, second in the world in my age group, right? For the marathon. So then I started to think, but how did that happen? Right? Like, what happened? And I had this realization, which is what triggered the book. You know, I hadn't gone faster than 243 in my 30s, not because I didn't have the talent, but because in some deep level, I didn't want to, right? All I had wanted to do was to be as fast as I was before I got sick. And so for that reason, I had stayed at 243 forever, right?

• 50:40Speaker 01

  I had like a mental block, or I like, it wasn't as though, I just, I never, it never occurred to me that I could go faster. I never set that goal because I just wanted to be who I was when I was 30. And so I broke through that. And so then what that made me realize is, okay, wait, the things that make you go fast, the things that make you go slow, they're buried deep inside, right? And they're buried deep in your psychology. And so that's what's led me to start writing the book. And so then I started writing about other people and their journeys in running. I started writing about my father, my relationship with my father. But the, the opening point, or the thing that inspired the book, was this realization that because running is such a simple personal sport, right? It's just you, you're just out there, it's intimately tied up in human decline. It is so personal and so intense that there's a lot there if you investigate.

• 51:25Speaker 01

  So that's what triggered the book. So it's called The Running Wall. And we'll see if it's any good.

• 51:29Azeem Azhar?

  It is. It's a wonderful story, and I'm looking forward to picking up a copy in October. The runner formerly known as Mr. 243, currently the CEO of Atlantic Media, Nick Thompson. Thank you so much for exploring this. This, we are going to see each other before the end of the year, and if not, in, in January.

• 51:47Speaker 01

  I'll definitely see you. I'll be, I'll be at all kinds of conferences. I'll be, definitely see you at Web Summit, who knows? I'll see you all kinds of places.

• 51:53Azeem Azhar?

  Thank you so much. Have a great day.

• 51:54Speaker 01

  Catch you later. Cheers.

• 51:55Azeem Azhar?

  Thanks for listening all the way to the end. If you want to know when the next conversation is released, just hit subscribe wherever you're listening. That's all for now, and I'll catch you next time.