Hiten Patel: Thank you for joining us on today's episode of the Innovators Exchange. I'm delighted to welcome Madhumita Murgia, the AI editor of FT [Financial Times] and author of her recent novel, Code Dependent. Thank you for joining us today, Madhumita.
Madhumita Murgia: Thanks for having me.
Hiten: Why don't we start with a brief intro to your role, both as a journalist and as an author?
Madhumita: Sure. So I've been a journalist since 2012. So actually I came from a science background. I was a biologist and an immunologist working on vaccine development, so had quite a shift in terms of career when I became a journalist. But I was really interested in communicating, you know, complex concepts, particularly the ones that I had studied and done research in. So I wanted to kind of be a science and health journalist to begin with, but my first job ended up at Wired Magazine, which is the Bible for tech nerds, and despite the fact that I was far from a gadget geek or tech nerd at all. I kind of fell into that world and fell in love with it pretty much in 2012. And because I was covering so much, that was the year, and those were the years when you really had the sort of rise of the mobile app economy, and you had all of these kind of companies that we use today in services like Uber and Airbnb, and so on. They were just tiny startups at the time. And I was writing about all of these great entrepreneurs and ideas. And so, yeah, I ended up being in the right place at the right time, and here we are. 12 years later. I've been writing about AI for a decade, about technology and entrepreneurship, and I've been at the FT 8 years, essentially covering and leading our reporting of AI globally.
Hiten: Amazing, amazing. And then talk to me a little bit about this most recent evolution and the book that you've just released to the space.
Madhumita: Oh, yes, completely forgot that. But so about 4 years ago I was actually on maternity leave, and I've been, as I said, you know, writing about AI for almost a decade at that point, and I could see that this technology was kind of moving really quickly. I've been fascinated by it for years, but I felt that there was a huge gap in the narrative in terms of what I was reading, but even what I was writing as a journalist. We often talked about the technology itself, kind of how fascinating the software is, the hardware, and also about the people creating it. But we weren't really thinking about the impact. So what's going to happen when AI hits the rest of us - consultants and writers and teachers and lawyers, all of the rest of us. How is it going to change our lives?
Madhumita: And for me, I've always really been interested in sort of people, and the human impact of technology, whether AI or anything else, you know. And so I wanted to write a book that focused on ordinary people, how their lives have been changed by AI today already, not in the future, but kind of what's already happening now. And I knew this was happening all over the world. So over the last kind of 2, 3 years I essentially found a dozen people across, you know, 10 countries, traveled around the world and really tried to understand how encounters with AI systems was changing and is changing the way that we live. And yeah, so really, this book is about the impact of AI and technologies like it on us as a society, on our work, on our kids, and on the way that we live, which I guess is becoming more and more urgent, as we are all talking about AI in our lives today.
Hiten: Yeah, it's an amazing piece, and let's get stuck in some of those findings in a little while. But before we go there, talk to me a little bit about how the transition from a news journalist to an author, what were you well prepared to do? As you embarked upon writing a long form book. What were the things that you kind of needed to bridge the gap on, what was new as you went into kind of a much more longer form, much more kind of thorough kind of piece, as you describe it, like, how did that transition work for you?
Madhumita: Yeah, I think, you know the big difference as a news journalist who's covered the area for many years, I still found that there was kind of a real shift. When you, when you're writing at a length of 80, 90,000 words, you can't just report on what's happening, or even just sort of the brief implications of why it matters. You have to have a much deeper thesis of change.
Madhumita: So you really need to think beyond what's happening to what do I think about it? What's my hypothesis for what this change means? And to look at it in in a much more sort of large-scale way compared to, I think when we report this in the news, it tends to be narrow, right? Like, why is this particular, you know, bit of news important for the industry?
Madhumita: What it means, maybe for the tech industry, or the corporate world, or a particular area. But not really how does this change the way we look at things? Or how does this change our work? You know I had to be much more sort of comprehensive and think more deeply about why I thought this stuff was happening, which pushed me forward as a thinker in this space as well I think.
Madhumita: And of course, writing 80,000 words isn't like writing 800 articles, right? You have to have a narrative. It has to have a coherent argument through from beginning to end, and it has to last as well. I think the great thing about news is you can break an amazing scoop, and it can have, like, you know, reverberate around the world for a few weeks. But most new scoops will die out within a few weeks. Then you're only as good as your last story. Right? You've got to go on to the next thing, and that's part of the joy of it, because you're constantly moving forward. But a book, it encapsulates a moment, or an era or an inflection point, and the point of it is that it's going to stay on and be relevant, no matter what happens in the space.
Madhumita: And it should still be something that people can read in 2 years' time and use what's in the book to grapple with what they're going through in the present. That was my goal, right? Like that this book has an impact that lasts beyond just an AI craze for today. I mean, I started writing it 3 years ago, before we even had ChatGPT. So the goal was never to just capture a hype, or a trend. It was to have a lasting, as I said, kind of a thesis about, how is this changing our lives, and so that, you know, to do that, I had to think about how I need to report differently, and what this book is going to look like structurally as well. And you'll see it's kind of structured going from the individual all the way up to, sort of, you know, when I look at China, it's at the national level. How AI is changing us, you know at all, from the individual to the collective. And so, I hope that we can use these stories and what we learn from them to kind of pass what's going on now with AI in our economy.
Hiten: No, it's incredibly well written, I must admit I'm only halfway through it. Yeah. And for those listeners that haven't read or are not familiar with your book Code Dependent? Are there one or two anecdotes that you'd particularly call out and draw upon that are kind of, that are pertinent now or stand that test of time lesson that you described earlier?
Madhumita: Yeah, look, for those who want to read the book, it's as you say, it's stories of people around the world. So, you know, I spoke to a doctor in rural India who's using an AI system to help diagnose tuberculosis on a really remote rural tribal population. And you know, she talks about the sort of value of having a machine that can do that in spaces where there are no human doctors. And this doesn't just apply to rural India, right? Or the developing world. We have huge shortages of doctors and trained medical specialists here in the UK and in parts of the West there's huge access issues for people of color in the States, for example. And so you can really see how a technology like this can bridge gaps and create entirely new sort of market, but also really help people. So that's one of the sort of positive ways that I looked at, how AI can better communities and their access to care. But most of the examples I looked at, what I found was, that even if you had kind of good intentions when you're trying to implement an AI technology in a say, in a workplace or in a specific setting, it falls down it. It ends up having harmful consequences because of how we implement it. And I think that's a big lesson. So an example, you know, is in Amsterdam, I focus on the story of a single mother, Diana, whose 2 children were ended up on a list that was generated by an algorithm by the Amsterdam Mayor, which used forms of machine learning to basically predict whether those children would go on to commit crimes. So it's very much like, you know, like a minority report style list. It was essentially, they wrote letters to parents saying your child is on a list, they're going to go on to commit a high impact crime, we're going to send in social services to help you out.
Madhumita: But the idea itself is kind of crazy and out there, but the goal was to help these families to prevent these boys from going down. It was mostly boys, by the way, who ended up on this list, and the majority of them ended up being from immigrant communities in North Africa. So even though these AI systems are supposedly gender blind and race blind, you actually end up, when you have these data systems having very, very targeted outcomes in this case, targeted at young men, boys, and at African immigrants.
Madhumita: But what happened there in practice, even though they were supposedly trying to help these families is that you know police knew these boys' names, they would kind of put a target on their backs, they would constantly bring them in for questioning. It made these boys act out more. It kind of tore these families apart. And it caused so much more instability, and fracturing, and kind of social chaos than if you hadn't had these sort of algorithmic lists.
Madhumita: And we're constantly doing this now. Even with generative AI systems, we're introducing them into HR, we're introducing them into creative jobs, into schools. And we're not really thinking about what it means to have a system alongside human experts and what that should look like. So I think one of the big takeaways for me from my book is about, how do we preserve human agency as we increasingly automate things that we used to do? You know? Maybe in your area. It might, you know, you have systems that might perform at par with a junior consultant, right? Or in my business, maybe at par with like a very new graduate reporter or graduate trainee in a law firm.
Madhumita: But what does that mean when you don't have people doing those jobs as much anymore. What does that mean for their development? For the health of the firm, you know, for kind of a pipeline development of talent. Like all of these questions we aren't considering, as we think of it, as a sort of easy or like a cheap fix, because it just feels very magical and kind of cool to use it right? So I think that's the big thing. How do you preserve human agency? So that we kind of continue to respect our own expertise of the people who've developed experience over decades in our various workplaces and ensure that we can hold these systems to account when they go wrong, right? Because they're far from perfect. They're just statistical systems. They are bound to go wrong. So who's holding the pen at the end of the day. Who's accountable for when they do make mistakes.
Madhumita: And you know we don't want to end up in a system like you know, where, like a, you know, a co-pilot system where we don't know how this stuff works, and so nobody can hold it to account. So that's the kind of big idea, I think, from the stories in my book which we can apply to what we're doing today as well.
Hiten: Such a powerful and real example. I think it really sticks and lands. What's the response you've had so far to some of those stories that you've put out there from the book. How have people reacted when you've taken this out to people?
Madhumita: Yeah, I've been really heartened because I think, you know, a lot of people have picked it up, having seen my work in the FT. And assuming it might be kind of a business book, or a sort of the good and bad of AI kind of book, and maybe been sort of surprised by how narrative driven it is, and how much about people. But I've, you know, talked to sort of lawyers and people from cross industry, you know, from BP, to banks, to private equity. For them each of the stories has spoken to some aspect of how they think about AI. And that's why I think people's stories always land, right? And they always speak to you, no matter what sort of background, you know whether you're a CEO, or you know, or a student, or whatever industry you're in, because we can all relate to just, you know, being a person. And who's having these unexpected encounters. And so, I've loved how people from all of these different kind of industries have felt that they can relate to these stories and have wanted to discuss. I think it's also a discussion, you know, it is about the ethical aspects of AI, but without just waving your hands in the air and calling it AI ethic.
Hiten: Yeah, yeah.
Madhumita: It’s like really in practice, what does it mean when you have a machine decide if your child is sick or not, like are you okay with that? Would you want a human doctor in the room? These are questions we really have to ask ourselves. And so I think that's what it's kind of prompted people to do. Lots of great response from policy people as well, like people in the UK Government. I had a great sit down in Washington with a bunch of people from the Hill, chiefs of staff, senators and others who are kind of really interested in what they could learn from these stories about policy making. Because this is what they're trying to think about. How is this affecting our constituents and our people, and they were kind of shocked. They hadn't ever come across real life examples in this way. So it’s been really interesting, the spaces that it’s gone to.
Hiten: Yeah.
Madhumita: Still curious to hear how Silicon Valley will react to it. I think there's so much optimism there around the technology and I think it's easy to see this as a sort of pessimistic take. But I'm not at all pessimistic, actually, about the outcomes of AI. I think I'm just more realistic about it. So yeah. So I think that that's an interesting sort of conflict there.
Hiten: I think it was a masterstroke to anchor it in the human stories. As you say, it's kind of counterintuitive. If you pick up a book about AI, everything is good about the technology and the systems. And to start and focus on the humans. As you say, that means it's powerful, and it brings out that lens. I think the other thing you've just laid out for me, which really resonates is there are so many other actors in the world, whether you're an investor, whether you're an advisor, whether you're a policymaker, whether you're just a citizen of society, we're all trying to get the heads up the curve and understand this, right? And some of these topics are just for the hobbyists. When we have people on who talk about digital assets and crypto. There's a version of you're in the club, and you're in the niche, and you love it, and you can kind of be part of that crew. But for me this is a topic that so many people are going to, either actively or passively be impacted. They need to kind of understand and take a take that it's really helpful then, to have narratives like that out there that are human driven, because that's probably the first way they'll think about it rather than this whole kind of talk to me about chips and the rest of it right? There's a lot out there. So I think it's yeah, it's a really important angle to be bringing to the table. Just let's go dig in a little bit more around just the mainstream narrative that's out there, and how you view it. Are there other key parts that you think are missing, or that's underrepresented, that you think people should be paying more attention to when they think about AI and the implications it's having on the world at the moment?
Madhumita: Yeah, I mean. So I think at the moment, right, there's just a lot of information overload when it comes to AI. So I worry that there's going to be some sense of fatigue around it. Where you, where people just sort of feel like it's out there, I don't really know what I'm supposed to do with it, and you almost sort of start to block it out. I think that's the mistake, right, because it's very easy to just continue your status quo and be like somebody else is going to deal with that stuff, and I'll eventually see it when I do. But I think the way to really engage with it, I'm not saying you have to read every news story or be fully informed but is even to just play with it yourself. And I had to make myself do that too, even though I talk about it all the time. And I'm thinking about it often but using it for a problem you have, whatever it is a creative problem, thinking through something, nothing is more visceral to help you figure out why it works and why it doesn't until you use it yourself. And also to criticize, if you want to criticize it, or if you want to push back, you still need to know what, how it kind of, what it does and what it can and can't do right. So for me, the first thing is like, even if you're so overwhelmed, and you don't know what it is, if you have access through your work, or even, you know, there are so many free ways to access these chat bots. Just play with it, you know, if you have something you're thinking about, whatever. It could be something, something silly like, you know, designing your kid’s birthday party invitation, or it could be something more than that, like, I don't know. I was brainstorming book subtitles on it for a bit, or like putting in a transcript of something, and then saying like, what are the big, summarize this for me. What are the big takeaways and trends? I found it like just an interesting way to get, you know, to get to go.
Hiten: Talk to me a little bit how you and if you use it day to day, and both as an author, and as a journalist.
Madhumita: You know, what I would say is like having played around with it, I am still one of those people, and I think it's probably because my job is so closely related to what a language model does right, which is generate language. That I still feel, like many journalists, resistant to it, and like there's no way it can do what I can do right? That's what my job is. That's what I'm trained to do, like, it's like, I guess, a form of like human arrogance. But you know, but also mistrust of it. So I've never, ever thought of using it to write a story, for example, or do the thing that I love about my job. But I have, you know. I don't use it every day, but sometimes I might do something like it's the kinds of things that search might not do very well, so like, you know. Tell me about this person. What do I need to know about them? Or how are these two people connected? Have they ever worked together? Do they have similar investors in their companies, you know, just to kind of, you know, map out relationships. It's really good at that. It's good like, if you're meeting somebody to get a really good, I was writing a briefing document about quite a big AI CEO, who was coming in to speak with us. And I was like, okay, I know this person quite well, but I'm going to see what one of their products essentially can tell me about them, and if it's good enough for me to do a briefing for my editor with it. Right? And it's, you know, it's as expected. It's basic. It's, you know, good enough. But that's, this is how I frame it in the book. I have my final chapter looks at sort of creativity and generative AI. It's good enough.
Madhumita: So it's a good starting point, right? But it's never going to be as good as the best of us, as good as the best consultants or the best journalists, because what you bring to it, what, you know, what makes you so good at your job, right? Versus somebody who's average or more junior. Right? It's, you know, it's the insights that you bring, because you know how to connect the dots. It's the surprising things that you say. It's being counter intuitive. It's revealing things that somebody else hadn't seen. And that's the kind of human reasoning and creativity that AI just doesn't have, but it's a good bare bones starting point and that’s quite helpful.
Hiten: Yeah. It's similar. I remember just I've been trialing it in our day job, and it's kind of hit and miss. There are moments when someone asked me a question like, oh, Hiten, you're an expert in this area, what's your view on that? And I'm like, well. I have a hunch, and you put it into the AI, and it's like, oh, here's three reports that validate the hunch. Oh, that's fantastic! And then there are other times when I'm like. I really want to get to the bottom of X, and you write a question like, it doesn't make any progress on that. So I feel like it is, as you say, you've got to tinker around, and I think personally, I found it a little bit hit and miss, times when it's really exceeded expectations and then times where it was like, actually, this doesn't seem to be the right avenue of pass. So it is clearly a point of experimentation, and clearly can potentially do some exciting things.
Madhumita: Yeah.
Hiten: Talk to me a little bit about, kind of, there is a lot of focus and attention on kind of the economic implications and benefits. Can this thing make money, right? There's a big new scrutiny. Now, if I think over the last year, there's been this endless amounts of focus on how much CapEx and investment companies being raised, spending money on chips. It was all about the input. I think we're starting to see a bit of an inflection point now around, okay. Are we getting a return? Is it generating revenue? Talk to me a little bit about that. From what you see.
Madhumita: Yeah, there was a Goldman Sachs, like a report that just came out. I think it was at the end of June, and I think it was something titled like, you know, "Generative AI: too much spend, too little benefit". Which kind of sets up that question, right, like a year and a half, two years ago, it was like, here's how many billions of dollars this is going to generate for the economy. And here's how many jobs we're going to replace. And you know, looking really at the value of it, which kind of drove everybody from, you know, from boards to CEOs to say, you know, we need to find a way to make this work in our IT budget. We need to pay for it. We need to play with it. But we're now sort of a couple of years into it. And it's not clear yet if anybody at all is making money because of it, other than obviously obvious players, which is the sort of hardware infrastructure players like Nvidia. OpenAI is making money. I wrote a piece a few months ago about sort of that their revenues are going like that [upwards]. But I think it's because, you know, they're in a period now where everybody wants to try out their tool. But I think it will have a window, right? A minimum window where people are going to see, like, is this, am I seeing any actual like impact on my bottom line? Or am I going to stop paying for this next year, or next 2 years? So I think they have a window to sort of prove themselves. You know the Bull Case, which is what you'll hear from Satya Nadella, and like the people who are building these models. And this technology is, you know, it's just going to change everything. It's like the Internet, you can’t quantify it. You can't say, oh, it's changed. This is what it's done to my revenues or my bottom line. You've gotta wait it out and go all in. But I guess on the other side like you're, if you're not seeing dramatic gains straight away, you're going to have a lot of sort of industry starting to think I'll come back to it when there's more here and like I'm not going to, you know, I don't have enough in my IT budget, to have just people playing around with ChatGPT. But you know, but I guess it's unfair to say we haven't seen the benefits yet, because it's so early in this space. It is like the Internet in the very early days, where everyone's just trying to figure out like, what can I do with this? It's like a powerful clay.
Madhumita: But what am I going to mold it into? And I think, you know, you are seeing a bunch of really interesting companies now. So that's what I'm kind of interested in focusing on for the second half of this year and going into next year, which is like what's happening on the application layer, right.
Hiten: Yeah.
Madhumita: What other? What are the areas and the industries where we're going to see real innovation on the application stage with when it comes to AI. Is it media, you know, like synthetic media? Is it going to be law firms, you know? So I, I'm actually working on something on robotics at the moment, which is cool. So yeah, I think that's what we will be figuring out over the next few years, and the jury is out.
Hiten: That parallel to the Internet era, and the need to be patient definitely resonates. I remember, like in the late nineties, we were going to the computer center at school. And everyone's obsessed about dialing onto this Internet thing just to see pictures of Formula One cars, I was like, wow, is this really it, like you can get this in magazines, guys. Like you could get this much quicker. And then I guess, a decade later, you can't live without it. But it does feel like it's a, there's patience versus kind of, and it may just pop out in ways that you can't predict in a nonlinear path for eyes. It feels like there's some of those parallels people draw with the Internet kind of resonate.
Madhumita: Yeah, for sure. And I think, you know, obviously, with VCs in particular, who need to see some kind of returns. I think there's a lot of like innovation you can do on the application side without waiting for some major breakthrough in the technology, like you don't need to have AGI to see some value today in terms of whether you're cost cutting. But also, I don't know, like improvements. It seems like customer service is an area that has seen genuine improvements. Klarna said, a few months ago, that you know they were, I don't know, it was like 20% or 30% up in terms of resolving customer complaints using AI. So you're seeing some real benefits. I know I spoke to Salesforce, and they had been working with Gucci, who said, like not only was like customer service, usually improved in terms of like actual outcomes, but the same people who are doing sort of resolving customer complaints had ended up becoming sales agents with the help of AI. And so we're actually acquiring new business and like selling when previously they were just there to sort of solve, you know, like sort of delivery issues, or whatever. So it's able to sort of augment the most experienced workers in a really interesting way. Which I think we'll see more of.
Hiten: That's pretty cool. That's pretty cool. One final one for you on this side. I guess we've had quite a few people on the show who kind of used to run specialist media publishing businesses, have kind of pivoted themselves into data businesses, because that's where value creation is. As a content creator, as someone who produces written content either as an author or a journalist, kind of where do you see the future value going here, is AI helpful or a hindrance? I guess we saw things like some of the big licensing arrangements between New York Times. Some of the AI models, I think Wall Street Journal have signed a similar one like, how do you see what you do in terms of producing written content and ideas being shaped by this over the coming years?
Madhumita: Yeah. So the New York Times hasn't signed a deal. They've sued them, actually, haven't they? So they've sued OpenAI, and that will play out in the courts. But the FT has signed a licensing deal as has the Journal and several others. So yeah, I think you're seeing different ways in which media production is thinking about how to take on this new threat, right, like, should we just join up with them? Or should we fight them?
Madhumita: So, but in terms of my work itself, you know, I think it's very easy to use this to automate some of the basic stuff that people assume. You know journalists do, or writers do. But I think if you think about where you get most value, the news that you love to read, the analysis that you love, the voices that you value. Why do you read that stuff? It's because it tells you something you didn't know before, or it surprises you, even in an area that you work in or know well. You know it helps you to, as I said, connect the dots right. Contextualize your little place in the world and show you why it matters, a deal you might have worked on, you know, suddenly, you're like, oh, this is how it fits into this wider ecosystem. Here are the other players, sparking ideas. That's what good journalism does right. It tells you, it informs you, even if you're an expert in your space, even if you're the CEO, or the Prime Minister, or the US President, we would hope you're reading the FT and you're learning something right. And that's something that I just, I don't see today's AI doing. But that's not to say it could never do it. I'm quite realistic about that. I think there can be advances where you know it's able to write, maybe even kind of put bits of news together to connect the dots. But I think actually, the media landscape will place more value on human specialists. And so I think the new differentiating factor will be, you know, our facts are vetted by humans. We have these reporters who are experts who are going and doing this stuff. We don't use generated media to do it.
Hiten: Yeah.
Madhumita: I think it will be almost like a...
Hiten: Premium.
Madhumita: A premium, I think so. But yeah, I mean, it's obviously going to be super helpful as a tool. I think, to like research, to report, you know, to put things together. We're already doing some really cool projects, using sort of visual AI and stuff like that. So I think there's opportunities for creativity. But I'm not sure that anybody wants to read a whole book just written by AI. I don't know, unless it's a textbook. Maybe that would be a useful way. Something to be entertained, right. Like, why would you want to read an AI generated thing.
Hiten: Yeah, hearing you talk to me, it just punctuates the difference between the production and the curation. Like, you can industrialize the production. And, as you say, it's not like we're short of things to read. Listen to and see right. There's just tons of it out there, but there seems to be like a real value, as you say in the curation of this is what we think you, whatever position you're in, like every man on the street or the CEO should be looking at. I think it becomes even more valuable for me, the curation when you're just drowning in. I've got five minutes on the way in, what do I need to know? Kind of as a selection.
Madhumita: Yeah, no, that's really true. I hadn't thought about it that way, but I think we are going to be sort of drowning in even more overload than we are today, because there's going to be generated video, images. It's going to be hard to know what the difference is, right, between real and AI-generated. So there's huge value in being able to differentiate and curate that, you're right.
Hiten: I'm just going to switch gears slightly. We always like to get guests who come on the show just to reflect on their, they've often had really interesting and unique paths, and yours is definitely up there. Given everything you've focused on, I'd love just to hear from you around any particular challenges or lessons learned over the past decade or so, as you've pivoted around that you think listeners would benefit from hearing about.
Madhumita: Sure, I mean, I guess for me, I'm now covering something that is constantly in the news. People of all kinds are really interested in it. And there's this hype around it, and it's great to be in the middle of a story like that. But I guess for me the lesson of like, how you dominate in this kind of space. So you stay on top of it is to basically follow what you're fascinated by always.
Madhumita: And you're going to end up in a place where this matters like, if you can figure out what are the areas that I'm really excited about that, I want to follow and understand really well, and get really deep into. I feel like that pays off. So I'm less of an advocate of sort of superficial jumping around between a bunch of things. And I've always wanted to kind of get really into something. I'm never going to be an AI scientist or an expert. But enough that I can kind of feel like I really understand the space. And then, as you follow that that turns into a massive story, and that's great. But even if it hadn't been this great story, I've just loved it so much, because I think it's such a fascinating space. And so I guess that's something I'll always do, like I never want to jump around into like, what's the next hot thing but like, just follow kind of what I'm really passionate about, and I think you end up getting the great stories from that.
Madhumita: I guess another big challenge for me, having covered tech over the last 10 years, it has changed so dramatically in the 10, 12 years, that 12 years that I've covered it, right. It went from being like a really fringe sort of small scale. Really, I mean, we've always had big tech companies, you know, like the Oracles and the Xerox, and whatever, Blackberry. But the mobile space that I covered it was very much like an innovator space. It was, you know, like accessible, and, you know, disruptive and scrappy. And the 12 years that I've written about it, these companies have become conglomerates you know, massive corporations, and the story has changed so much. There's now regulation is a big issue, like anti-Trust, but also it's become harder to cover them because they don't give you, it's not the same, as you know, writing to a startup entrepreneur, you know. I remember writing to Travis from Uber, who was just replying to his own emails like, yeah, sure, let's meet up. And now you have these armies of sort of you know, you have lawyers and PR professionals, so it's much harder to do the job. And so, I guess, like for me, the takeaway there has always been to be, you know, be as accurate as I can be, and fair in my reporting, fair and objective. And then, if companies like want to talk to you, great. But I found that it's always worked. If you're, you know, genuinely putting in the time and effort to be accurate in what you're saying, to build up your expertise in a space by talking to people, then even these companies who are, you know, they're constantly, people are banging on their doors to talk to them all the time. I find that I'm still able to kind of do that job just despite the challenges. But yeah, it's a very different world to the one I started covering in 2012.
Hiten: Fascinating hearing you describe how the individual stays constant, but as the company grows and evolves, then the persona, and the access, and the narrative, you know, as you say, go from replying to your own emails to being probably PR trained to within an inch of your life. I think we notice that sometimes even our own client dialogue when you've got founder CEO still in seat, the boldness, the clarity, the freedom of the vision, and how they express themselves versus, if you've got someone who is come up through a more corporate hierarchy, you know, big general, big corporate, then it is. Everything is generally kind of PR trained to within an inch and kind of, you know, it's hard to get anyone off script or say anything.
Madhumita: Yeah. And I feel like, if anybody you know in that world wants to say anything honest, you've now got platforms that you can do it directly. You know you can.
Hiten: Yeah.
Madhumita: You know you can go onto Twitter, or if you're really young founder, you could be on TikTok, or whatever, you know, like you don't need an institutional, like, news publisher to do it. So it's also, so much of it is building trust and like kind of saying this is why, this is who you can speak to if you're going to talk to someone like the FT. And this is why that matters, which you know you didn't really like. It was, that was different 10 years ago. So it's been like an evolving and interesting relationship with the tech world I think.
Hiten: That's amazing. That's amazing. Final one from me. We always like to invite guests to throw or share the spotlight. So is there kind of an individual or a company that's impressing you right now that you'd encourage listeners to kind of look up or pay attention to?
Madhumita: That's a tough one, because I feel like I write about all of these guys, and it's hard to pick one particular one. But let me see. So in terms of areas. So, as I said, currently, my current interest is robotics and kind of how that's been changed dramatically by the huge advances in AI, and I think Europe in particular has been like, is a huge innovation hotspot, for we're calling it now embodied AI, so kind of putting AI software into machines and getting them to do things that 5 or 6 years ago just would have been way too expensive, or way too slow. So I would say, you know, a more generic level to look at the really interesting startups like coming up across Europe in robotics, because that’s going to be big strides there over the next few years. In terms of individuals. I wrote a piece last year about the transformer. So this is like the core technology that has made large language models possible right? And it was essentially a paper that talked about the transformer model technology, was written in 2017. So it's only 6 years old. And it was done at Google by a team of 8 researchers. And they're all really different. I wrote this piece where I tracked 7 of them down. They've all left Google now. They're all doing amazing different things, and to kind of talk about how they came together in this moment at Google to kind of invent this technology that's changed everything. And one of them, Aidan Gomez, he now runs Cohere, which is a really interesting, really fab startup that builds LLMs. He was an intern at the time. He's an underground.
Hiten: Wow!
Madhumita: All the way up to Jakob Uskoreit and Noam Shazeer, who were much more sort of Google veterans who had been around much longer, and Noam now runs Character AI, Jakob runs Inceptive, so I guess what I would say is like check out that paper. Look at the 8 authors there, and look at each one of them has started a new company. I think just one of them is at OpenAI, but all of the others have actually founded their own companies and are doing new things in totally different areas. And I would say, you know, look, check out, you know, follow what they're doing. Ashish is another one. So yeah, they're all also from different countries all around the world. So it's amazing how international they are and what a different mix of backgrounds and experiences they had too. So I'm really fascinated by them as a group.
Hiten: That sounds awesome. Going to definitely check that out. Madhumita, thank you for being so generous of your time. Congratulations on everything that you've achieved today, and really excited to see what comes next. I think it's been a fascinating tour from Wired journalism into writing the book and some of the stories. I think it's great for us to be able to hear from you. So, thank you for coming on.
Madhumita: Thanks for having me.
