Inside how The New York Times leverages AI

0:00Starting a business comes with its share of ups and downs, which is why staying true to your vision is essential. a non-negotiable for Romeo and Milka Bregali, Capital One business customers and co-owners of Ra's plant-based restaurant in New York. Romeo and Milka took a leap of faith when starting their own restaurant, gutting an empty space and building it from the ground up. Every pipe, every wall, every detail. But building from scratch came with a heavy financial burden, which is when they turned to their Capital One business card. With the flexibility of the card's no preset spending limit, they were able to spend more and earn more rewards while bringing their vision to life.

0:36Today, Raz's success is proof that with passion and the right support, it's possible to make your dreams a reality. Learn more at CapitalOne.com slash business cards.

0:51I come from a family where the intersection of journalism and tech really kind of runs in our blood. And one of the people who I just admire the most is actually my great-grandmother. She founded one of the very first TV stations in the Midwest in the 1950s. She was definitely the only woman in any of these rooms and in any of these contexts. And I remember her describing to me that one of the first things she had to do to really bring TV to South Dakota and to North Dakota was to erect a chain of microwave transmitters. It was one of the longest chains of transmitters at the time that existed in the country.

1:27And so I think of her as, in many ways, the original technologist and kind of media mind. And she just had this sense of building and making to help figure out how to help support the business of journalism. And I remember thinking, like, that is what I wanted to do when I grow up.

1:46So that's exactly what Alex Hardiman did. She's now chief product officer at The New York Times, continuing the legacy of her grandmother. It's probably no surprise that with Alex's interest in technology and journalism, AI is a big part of her role. And while we've witnessed the rise of AI-generated articles on some platforms, that's not Alex's ammo. We really do feel like AI is an assistive technology. We are not using AI to replace journalists and to create journalism on its own. So how is the trusted newsroom of the New York Times actually implementing AI on a day-to-day basis? On this episode, we're answering that question and many more.

2:33We'll get into how the Times is using AI to build audiences, the future of journalism, and why Whirlle is still as relevant as ever. I'm Rana El-Khalyubi, and this is Pioneers of AI, a podcast taking you behind the scenes of the AI revolution.

3:00Hi, Alex. Welcome to the show. Thank you so much for having me, Rana. I'm a big fan and admirer of you and the AI conversations that you're helping to unlock for so many of us. Well, we are excited to have you on and we're going to talk about your role as chief product officer at the New York Times. But I just wanted to let you know we are avid subscribers and big fans of the Times. My kids who are 22 and 16, we have a WhatsApp group and we share articles and debate them on WhatsApp very passionately. And my daughter loves the cooking. I'm not a good cook, but my daughter loves the cooking section and she draws a lot of recipes from there.

3:36So we're big fans. Thank you so much for saying that. I mean, we try to make something for everyone in the course of their daily life. So that means a lot. Thank you for sharing. And then I have to ask you this, because at some point a couple of years ago at my company, Affectiva, we had a Slack channel for Wordle. We were all doing Wordle every single day. But then I don't know, I don't do that anymore. Do people still play the game? Oh, my goodness, yes. I mean, Wordle has had, it just has such cultural relevance. I kind of think of Wordle almost as one of our Taylor Swift strategies, because in the same way that Taylor Swift can bring together eight-year-olds and 88-year-olds from across generations, different ideologies, I really find that Wordle and the fact that people can come together to play the same game every single day and then share what was hard, their scores, what they're proud of, if they got Wordle in one, it just is such a bridging and bonding experience.

4:35And so it is still a very, very big part of our games portfolio and a really big part of what makes the times just a very community-oriented and culturally relevant product for so many people. Yeah, I love that. I remember having tons of conversations with people around their Wordle approach, right? Do you use the same set of letters every time or do you change it up? And it's very cool. What's your starter word? Do you have one? I do not have a starter word. I feel like that would be so boring. I don't know. Do you have a starter word? I used to. And then Tracy Bennett, our editor, actually said, you know, if you actually want to win, the best way is not to use the same starter word.

5:15It's to just get inspired by any word and then build a strategy off of that. So I definitely try to get inspired about whatever the thing is in my mind in the moment. I love it. So you're the chief product officer at The Times. And of course, the role of chief product officer is very common at tech organizations. But how common is it in a news organization? And what does the role entail? It's a really great question. So I am technically the first chief product officer. But when you think about one of the best products that The New York Times has ever made, it's the newspaper. So we didn't have a chief product officer, but our history of 175 years goes back to us knowing how to create this incredible corpus of diverse journalism to meet a range of news and life needs and to wrap that journalism in an extraordinary user experience that people can use to understand and act on the world around them.

6:08And so day-to-day, what that just means is I've got a portfolio of products that I help oversee in news, in games, in cooking, in sports with The Athletic, in shopping with Wirecutter. And we really were in these giant spaces where we believe that we can be truly the essential subscription to help people understand and engage with the world around them. And it's a privilege to be doing this job. Yeah, that's amazing. So I imagine your team consists of a number of product managers. do they need to have a background in software, a background in journalism, both? Like, how do you bridge these two worlds?

6:46So one of the things I actually really love about, I'll get to the talent point in a moment, because something that's really unique about the New York Times, and I was at the New York Times, by the way, for 10 years. Then I left and I went to Meta for several years, then went to the Atlantic and then came back to the New York Times. And what I'll just say, though, is there's something so unique about how we build product at the New York Times. We own the journalism. We own the software that helps to distribute that journalism. And we own well over 150 direct relationships with our subscribers and future subscribers.

7:21And so in many ways, when we think about building products, we kind of own the full stack. And that's very different than technology companies. And so the way that we think about building products is pulling together a product manager, an editor, a data scientist, a designer, a product marketer, a team of engineers. And it's the blend of art and science where we really have our editorial teams and our product teams truly embedded. So we can build products with a lot of creativity and magic and art, in addition to the traditional science that you see from a purely data-driven product company, more a la Meta.

7:56And so I'll say that product managers can come from so many different backgrounds. We see product managers who are former engineers, former designers, former journalists, former artists. And it's something I'm actually quite proud of for our team at The New York Times. Yeah. You talked about how the Times is kind of a full stack organization and you own the relationship with your subscribers. What does that mean? Yeah. So let me tell you a little bit about the business transformation of our story because this is a very, very central part of our current strategy. So if you go back about, let's see, about 10 years ago, when we were looking at our business, we were print first, we were advertising first.

8:42And to be honest, we didn't have a really clear path for growth. We had about a million digital subscribers at the time. Our newsroom was shrinking. And we were, to be honest, we were playing defense just to try to stay afloat. We weren't going on offense. And today we're in a totally different place. And it's been a hard one where we are a digital-first subscription-first business. We're now closing in on about 12 million subscribers. We've more than doubled our newsroom now to over 2 ,600 journalists. We have a business that is growing. It's a$2.5 billion business. But that kind of transformation, it did not happen by accident.

9:17If you go back 10 years ago, the media industry was largely, people were unbundling their content and their services. And the idea was to go chase traffic and ad dollars on the big tech platforms. And at that moment, we decided to essentially do the opposite, where we chose to be subscription first and destination first. And we decided to build proprietary technology and software that could prioritize those direct relationships so that we, not intermediaries, could really deliver trusted and differentiated value to our audience. and that was one of the most important decisions that we've made that still is a guiding principle to how we lead and build products and make decisions.

9:57And it laid the foundation for us to then make another consequential decision, which was to decide to become the essential subscription for, again, every curious person who is seeking to understand and engage with the world, and not just for news, but now also daily life. So this is your second stint at the New York Times, and after your first stint, you spent some time at Meta and it was a really crucial time for the United States and the world. So talk a little bit about that. I'd be happy to. So I first ended up at the New York Times almost 20 years ago, feeling like really one of the luckiest people in the world.

10:32And it was a really interesting period. I was there from 2006 to 2016 and it was a period of really deep transformation for the company. I was leading our mobile team and back in 2006, we had no presence whatsoever on mobile. And so it was really, really great experience that I then took when I went to Meta. It was 2016 and I decided to go to Meta to lead a team where we were helping micro sellers in low bandwidth parts of India come online for the first time and transact using social commerce and really looking at the interoperability of the Meta family of apps. And it was awesome work, but the 2016 election that happened.

11:10And as you referenced, it was a really intense time of reckoning internally, externally. A lot of really important conversations about misinformation, election integrity, and more were really taking place. And what I found myself doing there was attempting to do things to the Facebook platform that Facebook was not really built to do. Like, how do we classify and algorithmically rank accurate, high-quality information? That's just not what News Feed was meant to do. How do we reduce spam and abuse? How do we introduce fact-checking? How do we grow local journalism at the community level and really help bring people together at a time of really deep division?

11:52And so it was really hard. It was really interesting work, incredibly purposeful. But I remember thinking, I'm so passionate about the intersection again of journalism and technology. Journalism is not core to the mission of Meta. And Meta's been very upfront about that. Mark has spoken very, very, very clear. And so I, in my bones, I was like, how can I help bring these lessons in scale and ambition and product-driven growth back to the mission of journalism? And so that's when I ultimately found myself back at the New York Times in late 2019, right before the pandemic. I want to ask you a meta question.

12:28So at the beginning of 2025, meta canceled its fact-checking program. What are your thoughts on that decision? I haven't been at meta for quite some time. I can't comment on the specifics of what's happening there right now just because I'm not privy to it. But I do think in general, the need for veracity of information is only going to grow. And so when you look at kind of the bifurcation of the internet, you have all of this synthetic slop or dubious content, the provenance of which is increasingly difficult to identify, especially in a lot of these AI native platforms. But the human need to know what you are reading and that is accurate and that you can use that information to make really consequential decisions in your life, that is not a need that I'm seeing abate.

13:18I'm only seeing it grow in terms of its essentiality. But when I also think about what it means to develop a really high-quality information ecosystem, I do think that some of it lies with us as well. Like, trust in institutions has been eroding for decades. And we're seeing more and more powerful actors trying to discredit serious journalism. And so how can we come together to build an information ecosystem that is healthier, more accurate, and more trustworthy? And I just don't think that as an industry, we have done enough to make our reporting as accessible, as approachable, and as transparent as it can be and as it needs to be.

13:55And so when you look at our products now, we're doing so much to deepen how our journalists engage directly with the public. How can we make our work more approachable, more transparent? If you open the New York Times on a given day, you'll see in our app we have more reporters on camera who are speaking directly as experts to our audience about the process behind their stories. It feels like a friend who's just a very smart friend who's approachable and is saying, like, let me just tell you why this is such an interesting and important story and why it's one that hasn't been told but people need to understand.

14:27So as AI becomes much more ubiquitous and it becomes much more embedded across our daily lives, this work of building authentic relationships between our journalists and our audience and earning their trust just becomes so much more essential. In a minute, how the New York Times is using AI across their product portfolio, including in their recipes. Stay with us.

15:06If you've spent any time building AI products or leading technical teams, you know this. Transformation doesn't fail because of ideas. It fails because teams can't move together. Enter Atlassian's Teamwork Collection. It has planning in JIRA, documentation in Confluence, video updates in Loom, and now AI agents in Rovo, which connects the dots across your work so nothing gets lost. It's one AI-powered teamwork platform designed for how modern teams actually build. Learn more at Atlassian.com slash TeamChanger. That's A-T-L-A-S-S-I-A-N dot com slash TeamChanger.

15:55So let's switch to AI. You're not just leading product for the times. You are also leading the organization to become an AI-driven organization. So I want to dig into that. There's this tension between using AI tools and being at the forefront of this incredibly powerful innovation, but at the same time doing it in an intentional and thoughtful way while protecting kind of the company's IP. So how do you navigate that while kind of embracing the adoption? What does that mean for the times? Yeah, so two things can be true at the same time, which is what you're really getting at. There's a way for us to have an approach for how we use AI responsibly, and I'll talk about that.

16:36And then there's also an approach where we can say we also believe that our intellectual property must be valued and respected. So let's take the first one. So I would say that broadly, we have used technology for decades to help during each big moment of transformation. We did this with mobile. As I mentioned, when I was leading mobile, we have done this with audio. When you think about our early investments in audio, which have really helped us, you know, enter people's lives and homes and earbuds. It's machine learning is something that we have been using in our journalism and in our products now for a very long time to power ad targeting, to power our subscription model and our dynamic paywalls, personalized recommendations.

17:21And now AI, just like any other technology, is something that we feel can be quite powerful. But the goal isn't to be first to market necessarily, always with every new technology. it's to be the best at using technology to amplify the impact of our journalism, to really meet our audience's most important news and life needs, and to solve real problems at scale versus chasing fads. And this is something that we really try to focus on because we could get pulled in a million directions. So there are a few principles that I'm happy to share that really guide our work and how we use AI responsibly.

17:54The first is that we really see AI as a tool in service of our mission. And that mission is to seek the truth and to help people understand the world. And Rana, there's actually something that I've heard you talk about on some of your previous podcast episodes about this idea that AI should be about augmenting and amplifying human ability. And ultimately, it can unlock greater human potential. And that's how we see AI, right? We see it as a tool to assist our journalists and to accelerate our business. The second, though, is that we always have a human in the loop. And this is really important.

18:27the expertise and the judgment of our journalists those are our competitive advantages and that's what machines simply can't match because if you think about it journalists are out in the world and they are finding information that does not want to be found to tell important stories so this is information that no llm has access to and i really believe that that is information that will even more important in the age of AI. At the end of the day, our brand promise is that we are always responsible for what we report, however the report is created. Yeah, I love the kind of the human-centric approach to harnessing AI, because as you mentioned, this has been the common thread in my entire career.

19:11And I think it is a combination of keeping the human in the loop and keeping the human accountable, but also thinking about how to use these tools to augment and amplify and tell stories that we've not heard of yet? 100%. And, you know, just to sort of give you an example of like what this looks like in the real world, there was so, I want to say it was maybe 10 days before the election, this was the 2024 election, two of my colleagues in the newsroom, Ali Brazon and Nick Caracentini, they were reporting on Trump's campaign election lies. And what they got their hands on were over 500 hours of video recordings of the group that was behind this movement.

19:51But the problem was there were fewer than 500 hours before election day, right? So like, even if we put the entire newsroom on the task, we wouldn't have gone through every single video in time. So this is where we were able to take two members of our AI team and embed them with a journalist to transcribe the videos. And then they fed the videos more than 5 million words of transcript into an LLM, which Ali and NIC using their expert judgment used to search and prompt and surface examples for their reporting. And this reporting was then able to be published in a matter of days so that people had access to this information before going to the ballot box to decide who they were going to vote for.

20:29So that's just, I think, such a great example where, you know, the LLM on its own doesn't do much. But when you have a high-quality data set and a set of expert journalists who know how to prompt it and ask the right questions, you can really unlock a lot of assistive potential that can really help make journalists all the more powerful. Okay, so my son is doing this archival research, looking through Egyptology archives, and there's thousands of documents in Arabic, and he's not very fluent in Arabic, but he's been using a variety of AI tools to essentially streamline some of that process and help kind of this professor he's working with to sift through all these archives.

21:08So I think what I'm hearing you say is AI is going to change a lot of the ways we uncover data and uncover stories and, yeah, accelerate some of this kind of research. It's pretty cool. I think just research in general is definitely a place, as long as you have a human loop to then fact check. Again, it just all comes back to making sure that you close that loop. But I do think that that's a very relevant and very prevalent use case. How do we support the craft and the making of journalism, but in no way is AI going to replace it? I love it. What are some other examples of how AI is being used in the newsroom?

21:46So I would say that at the story level, there are so many other examples of machine learning engineers being paired with journalists to sift through giant public data sets and to find and uncover really important patterns. But beyond that, there are so many other examples across our products. And so this crosses the newsroom and our product teams where we're really using AI to make all products in our portfolio so much more accessible to more people. And you might not know this, but we actually have generative AI now embedded in many of our larger consumer products. So in news, for instance, if you just open up the product on any given day, over 90 % of our news and opinion stories now have an automated voice mode attached to it.

22:29And so this is a very simple way for us to take a story that traditionally would have only been in text format and create a different modality so that if people are commuting, if they just prefer to listen, maybe honestly for other reasons, audio is the only way that they can access and retain information. It just unlocks from one original story all of these other high-quality derivative versions that could just travel more widely to reach and engage more people. Cooking is another one of my favorites. I know you mentioned that one of your children is into cooking. If you're ever outside of the U.S., historically, it was hard to cook with us because our recipes were in imperial units.

23:10And so we've been using generative AI to do very simple things like help automatically transfer ingredients from imperial to metric units. And these are small grace notes, but you can find that these types of things, when you really understand a pain point in your product and you can thoughtfully apply generative AI, we now have over 50 % of our international users using this feature and just finding that the product helps them cook in a very easy and effective way. So there's so many ways like that where it might not be the most splashy use case, but what it really is doing is unlocking the power of our journalism for so many more people.

23:46And that's incredibly valuable. And that's the type of impact that we want to see. Yeah. And it's enhancing audience engagement, right? Like back to kind of building this authentic relationship with your audiences. That's kind of one way to do it, right? Right. A hundred percent. And what we're finding is that for people who already have a relationship with us, it is helping to drive more engagement because they're just getting so much more value out of the product. But then for other audiences where there might've been a little bit of friction based on the way we were distributing our content, the way that we were presenting it, we weren't necessarily multimodal.

24:19The more that we reduce those barriers to entry, the more that they're finding their way into the New York Times universe as well. And as we take the long view to build the next generation of future readers and subscribers and listeners and viewers. It's a really important part of that strategy. Yeah. Were there some tools or use cases that were completely off the table? So the AI use cases that I think are off the table are the ones that in any way compromise the independence, the originality, and the trustworthiness of our journalism. So I don't have a specific example because in many ways the principle has kind of put just enough guardrails so that there's a lot of room for experimentation, but we're also not off exploring things that we would just never ship.

25:02So there are some pretty hard lines there about what we will and will not explore when it comes to the journalism itself. But beyond that, we have a really high tolerance for experimentation internally. So just as an example, a couple of weeks ago, we had one of our internal hackathons called Maker Week. And this is where we have so many colleagues from across the company. Anyone can basically come and play and use AI tools to figure out, you know, how to solve a problem that's been on their mind that they wanted to solve, how to unlock new value. And we ended up with over a hundred different prototypes leveraging AI.