Transcript: The Path Forward: Artificial Intelligence with Marissa Mayer, CEO & Co
MS. MONTGOMERY: Hello, and welcome to Washington Post Live. I'm Lori Montgomery, and I'm the business editor here at The Post.
Today, my guest is Marissa Mayer, here to talk to us about artificial intelligence, tech leadership, and more. Hi, Marissa. Thanks for joining us.
MS. MAYER: Thank you so much for having me.
MS. MONTGOMERY: So a lot of our viewers will know you, of course, as the CEO of Yahoo and the first female software engineer at Google where you designed some of the foundational tools of modern life, certainly of my life. I think I use Google Search and Google Maps pretty much every day, and I know you are at work on some new tools as a founder and CEO of Sunshine, and I want to get to that in a minute. But first, I wanted to talk about the current moment in Silicon Valley, which is so dominated by advances in artificial intelligence.
Help us think about AI. You said that you think that the AI will set the tech industry ablaze, and I think that's been undeniably true. Here in Washington, meanwhile, we hear a lot about AI as an existential threat, a technology to refer a current popular--refer to a current popular movie as something like the atom bomb, something that we don't fully understand and may not be able to control. How do you see it?
MS. MAYER: Well, I think, first of all, let me clarify the "ablaze" comment that I made when we were talking about was the current moment, which, of course, is really electrified by AI, but it's also defined by a lot of economic constraints, the downturn. You know, there's a lot of things right now that are really weighing on different tech startups and large companies as well, and so in that, what I had observed is there have been other downturns like this; 2000, for example, the financial crisis to some extent. And what coming out of those, I do think that it's a downturn, but it's usually followed by an upturn. I think AI is very much going to be the thing that propels us forward out of this moment.
And I think that what you often see, as we saw in 2000 and again probably in 2008-2009, is some of the strongest companies moving forward, the ones that really grow, really provide value to people's everyday lives, really get defined by these moments, because the strongest companies ultimately survive these types of downturns and harness the types of technologies, like AI, that can really propel us forward.
MS. MONTGOMERY: But will it kill us?
MS. MAYER: I--personally, I'm an AI optimist. I think that AI is going to really help us, both on an individual level as well as across humanity. I think there's a lot--going to be a lot of advances that happen much more quickly because of the assistance of artificial intelligence.
MS. MONTGOMERY: So do you think all this AI doomsday stuff is--we should ignore it, or how should we think about it? I mean, you know, every day it seems like someone else is speaking out to say, "Whoa, slow down."
MS. MAYER: I think that one of the most alarming things in the current moment is things are happening really fast. Technology is getting very good. Things are moving really quickly, and I truly think there are some pitfalls with AI. And we need to be very aware of some of the dangers, thinking ahead in terms of what could happen, how do we protect against that, how do we really take a human-centered view to artificial intelligence, and how we can assist humans as opposed to replacing them or destroying them. And so, you know, in that, I think that there's a lot of things to consider, and so it's certainly a time where we have to be very thoughtful. But I think it's something that ultimately is going to be looked at as a moment that was really revolutionary in terms of what we can achieve.
MS. MONTGOMERY: Okay. And you are a bit of an expert on the subject. You were drawn to AI at a very early age. I think you specialized in the subject both as an undergraduate and as a master's student. What fascinated you about it back in the day? Why were you drawn to the subject?
MS. MAYER: Well, I just think the way human minds work is infinitely fascinating, and I grew up thinking I wanted to be a neurosurgeon. And when I got to college, I realized I was less interested in the medicine of the brain and the structural analysis of the brain and much more interested in terms of what was happening. How do people learn? How do they reason? How do they express themselves?
And then there was this notion, especially at Stanford, because it's very strong in computer science, of can you make a computer that does the same, which I thought was fascinating. Can you actually replicate in silicon? How can you actually--you know, how can you have that really start to resemble human-like intelligence? And I think it's an amazing challenge. It takes a lot of understanding across a lot of disciplines, and ultimately, that's what really drew me into the area of, first, symbolic systems with a focus on AI and then, later, computer science.
MS. MONTGOMERY: Why exactly does this feel like an epochal moment for AI? What about our lives is about to change that we don't really realize? I mean, like Google Search, for instance, that you are such an expert in, is the way we learn about things about to really like dramatically be very different, or what's about to happen?
MS. MAYER: Well, I think that what you can see is--you know, in the early development of Google Search, we--it seemed very smart, but what in truth, it was just extremely exhaustive. We were able to take in all the information, really analyze that information, surface it in terms of overall relevance.
Now what you're going to see is the search engine really--and the way we do search, the way we learn about things really being able to take into account your preferences as well as answers that you think are good, that you think are bad, that you think are irrelevant. And I think you're going to--we're going to get to, especially in Search, an area that's much more efficient with an interface that feels much more natural to people.
You know, in the early days of Google, we found ourselves at times having to kind of help educate people. How do you put together a good keyword search? With the rise of LLMs and chat as an interface, all driven by AI, you'll be in a state where you can just ask questions in a way that feels very natural to you and have a discussion in a way that again feels extremely natural and intuitive. And I think that both the interface interaction as well as the quality and efficiency of answers is going to change remarkably.
MS. MONTGOMERY: And do you worry that--I mean, one of the things I love about doing research via Google is that I can get to original source material. Do you worry that we'll become--it will become a function of just getting answers spit at us without--you know, in a time when disinformation is such a concern, that people will lose that connection to where is the information that I'm getting actually coming from?
MS. MAYER: I think that disinformation is a concern. It's obviously one of the big things that people have brought up as a downside of AI. That's how I think that AI itself can actually be used, to hopefully find disinformation, and one of the ways that you prove that something's correct or incorrect is by going back to source. You know that obviously as a journalist, but it's not hard to imagine us being able to train technologies that do the same type of analysis that you do to really understand. Is this trustworthy? Where did it come from? Is it reliable and well researched?
And so I think that that notion of being able to get back to source--well, for the high-level questions where you could ask it. You're fairly certain it's right. You may not go all the way to source. But for things that are more involved and things where you feel you need more reliability, it's not hard to imagine having the AI search or whatever type of agent you're interacting with put a big premium on that source material and perhaps even produce it or point to it.
MS. MONTGOMERY: Yeah, I just worry that my mother already can't tell me where she learns half the things she reads. So I'm a little concerned about this.
But okay, now we have a question from my colleague, Gerrit De Vynck. Gerrit covers AI for The Post in our San Francisco Bureau. He asks, "What is your view on labor displacement? There's a popular idea that AI will augment rather than replace workers, but that seems too convenient and pat. Do you think we will see people lose jobs to AI in the next five years?"
MS. MAYER: I actually subscribe to the view of the first part of this question. I think, by and large, what we're going to see is an increase in productivity, in excellence, in consideration of even far-flung solutions and possibilities, and really using AI as an assistant.
And, you know, everyone from executives to artists, creative, administrative, people use a lot of different assistants. Most of them happen to be humans. We're about to enter an area where some of those will be computers or technology. So I think that most of the advancements we see, at least in the short term, the next three to five years, will be of that form.
That said, there's undoubtedly going to be some--there's going to be some job displacement. It just will happen. Almost every technology causes some amount of job displacement. The question is, what are the new form of jobs that come after that? How do people use the time that the technology has given them back? All of those types of things are really the questions that I think we should focus on.
But for me, the real thing is, what can we achieve with an assistant that can work like this? What can we achieve having that technology available?
MS. MONTGOMERY: Yeah. Well, I think those are the two big concerns we see here in Washington is, will it kill us and will it kill jobs?
Heading into the 2024 presidential election, I think we'll see increasing concern about how AI might supercharge deepfakes and fraud. We've also, in our reporting at The Post, written about how it could help hackers, how it's already being used to create realistic child pornography. What--these are all difficult problems, I know, but what should we expect from companies to do to start to limit some of these harms?
MS. MAYER: I think that, as I said, we can actually use the technology against itself. That's the most obvious thing to me to do is to actually use the technology to understand what's deepfake, to use the technology to understand some of the misinformation, and to actually put guardrails on the technology against creating illegal content or really harmful content.
And I think that the hardest part about this moment is that even experts in AI don't know where we're going to be in three to five years, and when you're talking about regulations, guardrails, you really need to understand not necessarily where the technology is today but where it's going to be three to five years from now. And some of the best people in the world would have a hard time predicting that right now, and that's the thing to keep in mind. So I think that we're going to have to see a real partnership between regulatory bodies and the companies themselves, right, companies really deploying the technology in a way that they think is additive, creates a lot of value for the user and for society as a whole. I think those are some of the guiding principles, and as I said earlier, I think it's really important to take a human-centered view where we really look at how can AI really move humanity forward and how do we put the humans really at the center of what we're trying to achieve.
MS. MONTGOMERY: So what role do you think government could play right now? You mentioned a partnership. I know Senator Schumer is supposed to be working on regulatory legislation. We haven't seen what that looks like. What is the appropriate role for government? What posture should Washington be taking?
MS. MAYER: I think they're doing some of the right things. I certainly think learning about the technology, asking the tough questions, asking for accountability from the companies who are making big advancements here is a starting point, and asking those critical questions also makes those companies think more carefully about what they're doing, what the ramifications could be on a broader level, et cetera.
On the actual like regulatory piece, I think I would caution that I think we need to take a cautious view. I think very specific regulation that's reacting to where the technology is right now is likely going to either be, probably in the best case, irrelevant to where technology ends up three to five years from now, and in the worst case, it could actually curtail it.
And so, you know, right now, I feel that in AI, there's a very U.S.-centric view at the current moment. It seems that the United States is ahead. There's a lot of other countries that are working on this, China, most notably, that will have their own set of views, values, regulations that govern this. And we need to make sure that the technology that we're advancing, that I would argue largely here in the U.S., is human-centered, doesn't end up curtailed by our own regulation in a way that's different than some of the international competitors.
MS. MONTGOMERY: Have you spoken to any of the policymakers here in Washington, Senator Schumer, or anyone else about this?
MS. MAYER: No, not recently.
MS. MONTGOMERY: Okay. I've got a similar question from my colleague, Cat Zakrzewski. She covers tech policy for The Post here in Washington. She asks, "The White House has hosted two major meetings with executives from AI companies. Those meetings did not include any female executives. Are you concerned that the gender disparities the tech industry has struggled with for decades will be repeated in AI, and if diverse perspectives are left out of these conversations, what implications could that have for the future of AI policy?"
MS. MAYER: Well, I myself am really heartened that some of the large companies in this area have prominent female executives. The CTO of OpenAI is a woman, and so we definitely are seeing, in my view, representation that's at least as good as the technology industry at large, if not somewhat better in terms of people really playing a formative leadership role in these areas. I'm really proud of what some of the women pioneers in the space are doing and the roles that they've played.
I think that what we are seeing is the current AI takes as its input data and training data, basically the set of human knowledge to date as it exists on the web, and so that technology will have in it all of the same types of biases that the current web has, that the current knowledge has, and that's just something to be aware of. There's some ways to correct for it, but overall, I don't think that there's an overall, in my view, sense of alarm. I think that it would obviously be better to have more women involved in technology and AI in particular. I see a real interest from a lot of my female colleagues and a lot of young women who are just entering the field, which is great. And as I also said, I'm happy to see women playing leadership roles across the industry in the AI space, and we just need to make sure that some of the biases, you know, gender-based as well as other biases, aren't perpetuated by the current set of training data, which reflects the biases of the moment.
MS. MONTGOMERY: Okay. Now I want to talk about Sunshine. Your website says that it's all about making the mundane magical. Explain.
MS. MAYER: Well, we founded Sunshine because we felt that technology should save people time and it should make you happier, and so for us, what that meant was saving time usually comes down to taking everyday tasks and making them more efficient, and making people happier usually comes down to relationships.
Harvard just finished one of the longest-running studies on human happiness ever. I think it ran for more than 70 years, and time and money don't make you happy. What makes you happy is feeling connected and having strong relationships with other people, and so we thought about what are some of the problems today that people deal with on a daily basis that could be handled more efficiently and ultimately make your relationship stronger. And there's a whole class of problems from contact management, which was our first product, to remembering birthdays, but also things that start to become much meatier, groups, events. How do people get together? How do they spend their time? How can we assist them with that? And what we see is there's a lot of really cutting-edge AI being applied to that bleeding edge of technology, but there's also value in applying it to everyday mundane problems and the types of advances that you get from that.
And I would also argue that by applying it to more mundane everyday problems, you also start to build a familiarity with the type of technology, with what it can do, and ultimately trust. So I think there's--we felt that as we started Sunshine, there was a lot of things that we could achieve that were particularly important, and it may not necessarily be as cool as self-driving cars, but it's something that we think affects people's lives every day.
So we started with two products, both of which are in the app store and one of which--and as in the Android store, there are Sunshine contacts and Sunshine birthdays, and we'll branch out from there, because ultimately, we want--what we want to help people do is build a system of apps and knowledge that really help them manage their relationships more and give them time back.
MS. MONTGOMERY: Did you start with the idea of how do we help people with relationships, or did you start with the idea of how do we increase human happiness? What was the original spark?
MS. MAYER: I will say that I think we started more with how do we give people more time back? What are the things every day that just take way more time than it should? We have global facial recognition. But yet if you watch what's happening, everyone is spending a lot of time pinching and zooming and making sure eyes are open and that the expression is the best expression of the set of pictures they took of their children or of their friend, and so you're like wait, like we can do this amazing stuff at the frontiers, but the kind of everyday thing of everyone pinching and zooming and looking at faces in a zoomed-in way is still--you know, takes an inordinate amount of time.
Contact management is so boring and so tedious. Nobody even does it, right? Every contact book we run into is some form of a mess. The question is, what form of a mess? And we can help most of them. But that happens because it's so tedious to clean it up. People just don't do it. That said, computers and technology are actually really good at solving these kinds of problems, and so we wanted to see these types of things applied. And in our--in the end, we felt that, you know, what you should have is a lot of companies in the modern day, especially in consumer tech, are motivated with getting more and more of a user's time. How many minutes per user per month can you have? How many minutes? How many active users do you have? And we felt that as technology gets good, it should start to fade into the background, and some of these things should just happen automatically. You should actually get time back, spend less time achieving the same outcomes, and that's really where our technology is focused.
MS. MONTGOMERY: So I love that. I could use more time. I could also use help with my address book, which is a complete disaster. But your website says that you're not going to sell ads and you're not going to sell my data. So how are you going to make money? Are you going to charge me for this?
MS. MAYER: That's right. So some of our products are so new that we're just overall looking at what value users get from them. But our Contacts product, for example, you can use it for free if you use Sunshine Contacts. But if you want us to improve your address book on your phone, your iOS address book or your Android address book, you pay us a subscription for export, where we take your contacts, clean them up, and export them to your phone. And we also keep them up to date. So as those contacts change, move, and so on and so forth, we keep them up to date in an ongoing basis. So your address book is always complete and enriched by Sunshine and up to date.
MS. MONTGOMERY: I would be a bad reporter if I didn't ask you what--how much would this cost me?
MS. MAYER: Oh, sorry. It costs $4.99 a month, and the first month is free for people to try it out.
MS. MONTGOMERY: Awesome.
I also love on your website how you sum up your company culture: Be nice or leave. This seems like a pretty rare attitude at tech startups. You've talked about how in the early days at Google, Larry Page and Sergey Brin just yelled at people until they did what they wanted you to do. Has your philosophy about management changed since then?
MS. MAYER: No. And I will say, Larry and Sergey were never mean about things. They were insistent. They knew what they wanted. They knew what they would consider a job well done. But you know--and they--there was an insistence to it, but overall, it wasn't unkind.
And, you know, it's funny. The "Be nice or leave" was a pillow that my long-time management mentor, Maureen Taylor, gave to me when I first got to Yahoo, because there were a lot of people at Yahoo who are nice. There were definitely some people around the company that were less than nice.
[Laughter]
MS. MAYER: And so she kind of gave me the pillow as a joke, but people were like this is great, right, because, you know, life is just too short to work with people who aren't nice and don't have the best of intentions.
MS. MONTGOMERY: Yeah. I'm with you on that too.
Okay. So a little bit more about management philosophy. You've become sort of famous for a hard-charging style. You only took two weeks off when you had first baby? Second baby? Both babies? Yeah.
MS. MAYER: Something, you know --the job had various demands that needed to be met, but yeah.
MS. MONTGOMERY: You talked about working 130 hours a week by managing meals and bathroom breaks. At Yahoo, you ended remote work. Post-pandemic, we're in a time when certainly remote work and even four-day work weeks seem to become--be becoming ever more firmly embedded in American life. Is this a mistake, in your view?
MS. MAYER: Well, I think that in each of those cases and examples you gave, I think it's really about being flexible and responsive to the moment.
In the early days of Google, those 130-hour weeks, we just knew that we had something that was incredibly helpful for people and incredibly special and really needed to capitalize on that moment.
When I came into Yahoo, it was a company that was in decline. There was a lot of energy from the people there in terms of how do we get it moving in the right direction and return it to greatness, and that the ending of remote work actually rose up from the workforce itself. They said, look, we have this policy that people can casually work from home when they want to, not in a formal setup, not in an established routine, and it's really getting in our way. And as I observed things, I agreed with that, and so I felt like it wasn't necessarily a comment on what's right for all companies or all people. It was just what was right for the company at that time, and obviously, in the pandemic, the safest and best thing to do was to work from home. And so we're very supportive of that here at Sunshine and across the world really. It makes--it overall makes sense.
And so I think you have to respond overall to the moment, and I would say right now at Sunshine, we're back in person. And I will say, overall, the people here, they like it. They like being able to work with their collaborators, but we're a small enough company that it makes sense to have everyone together. It really aids fast-paced discussion, experimentation, trying new things, moving forward on ideas, discarding ideas. That all just happens in a much faster way if you aren't constantly having to schedule discussions of things, the fact that you can have that kind of informal office discussion is really helpful, especially to a small company. And so it's coincidental that it all feels very hard-charging, but it's really responsive to the challenges and opportunities of the moment.
MS. MONTGOMERY: So anybody looking for a job at Sunshine should not expect a remote position.
MS. MAYER: That's right. Right now we're all here in person. I do hope that Sunshine--I think as companies get larger, we can become more accommodating of remote work, but when we're this small and we're still figuring out where do we want to grow, what types of products do we want to develop, these kind of formative stages and early stages, I think it's really important to be together.
MS. MONTGOMERY: So your years at Yahoo sounded pretty rough. You were, of course, already under a massive microscope as one of the few women CEOs in the Valley, first Fortune 500 CEO to give birth, first to be featured in Vogue, if I've got all that right. Are there things you would have done differently, and how has that experience shaped your decision to found your own company now?
MS. MAYER: I will say I'm very proud of my time at Yahoo. I think that the product portfolio that we evolved to taking Yahoo really from the web and making it popular on mobile and making it a really big mobile platform, I think was ultimately the right strategic move. And there's a terrific team there, and we were able to build a lot of value both for Yahoo users around the world as well as for our investors.
I do wish that it had been--we'd been able to really get it back to the growing behemoth on par with Google and Facebook, which was the hope, and so, you know, I would have liked to have seen a different outcome for Yahoo rather than a sale of the operating company. But overall, I think that it's--no. It's I don't regret my time there at all, and of course, you learn from every experience, and I made a lot of mistakes. I will say most of them is the bigger mistakes, when I reflect back, have to do with sometimes not taking a bold enough stance. It felt like we were taking a stance in terms of bold acquisitions, shifts in the portfolio. Some of those things, I would have done to an even more extreme, and obviously, hindsight is 20/20. You know where to invest and where not to invest now in the--in retrospect.
MS. MONTGOMERY: We're almost out of time. But I did want to ask you, you were part of a generation of women who broke down barriers in Silicon Valley at the big tech companies. Sheryl Sandberg is another example. Meg Whitman. The C-suites now, not so many women. What does that mean for women in 2023?
MS. MAYER: Well, I mean, I think that we'd love to see more women in leadership positions. It does seem--you know, I don't know if it's anecdotal or actually quantitative, but it does seem that there's a bit--a few--a bit fewer, but I do think these tend to come in cycles. And so I'm hopeful that with--especially with the rise of artificial intelligence, that we'll start to see even more women leaders. The people who were in college and training 10 years ago, 15 years ago are starting to step into those really important director, VP, and senior VP roles across the Valley, and those are ultimately what lead to the C-suite. And so I think that we'll continue to see that grow and evolve, and we overall need just to be supportive of it. And it all starts with the beginning of the funnel of getting women into technical fields and computer science, in particular, at early phases. Most studies show that a lot of the drop off actually happens in junior high or high school. And there's a lot of different organizations and people working on how do you change that over time, and the numbers in computer science have ebbed and flowed over the last 20 years, but my understanding is they're back on the increase now. And there certainly have been peaks over the last 20 years that I think we'll start to see the fruits from.
MS. MONTGOMERY: Okay. Well, I think we're going to leave it there. Good luck with your new venture. Thank you so much for joining us, and good luck.
MS. MAYER: Thank you so much.
MS. MONTGOMERY: And thanks to all of you for watching. To learn more about our upcoming programming, please go to WashingtonPostLive.com. I’m Lori Montgomery, and thanks for joining us today.
[End recorded session]