This transcript is generated with the help of AI and is lightly edited for clarity.
BRET TAYLOR:
We are incredibly creative, and I think we will create new jobs that will be high leverage, high impact. I’m really excited to be wowed 50 years from now by what our economy looks like and all the new jobs that will just be as foreign to me as perhaps podcasting would be to Benjamin Franklin, because I think it’s really exciting.
REID:
Hi, I’m Reid Hoffman.
ARIA:
And I’m Aria Finger.
REID:
We want to know how, together, we can use technology like AI to help us shape the best possible future.
ARIA:
We ask technologists, ambitious builders, and deep thinkers to help us sketch out the brightest version of the future—and we learn what it’ll take to get there.
REID:
This is Possible.
REID:
Hey there, Possible listeners. We have a special two-part episode for you today. At the 2024 Masters of Scale Summit in San Francisco, Aria and I sat down with Sierra co-founder and OpenAI chairperson, Bret Taylor. He’s been a leader at both startups and large tech companies, has both management and technical chops, and has been at the center of it all during a number of technology revolutions.
ARIA:
So, needless to say, Bret knows his stuff—both as a leader of startups like Quip and FriendFeed, as well as former CTO of Facebook, former Co-CEO of Salesforce, and former chairman of the board at Twitter. And we had so much more to ask him that we invited him back to continue the conversation so that we had a full episode’s worth of discussion. So today, you’ll hear that live show with Bret from Masters of Skills Summit, and then we’ll go right into our expansion of that conversation.
REID:
This was a fun one. Here’s our conversation with Bret Taylor.
REID:
Bret, thank you for joining us.
BRET TAYLOR:
Thank you for having me.
REID:
So let’s, let’s start with what’s your picture about what’s going on with AI, kind of generally in the world and then in Silicon Valley?
BRET TAYLOR:
I’m really excited about this current wave of AI. You know, I think in Silicon Valley we’re prone to trends, [laugh], and some of them—
REID:
I haven’t noticed them.
BRET TAYLOR:
Yeah, yeah. It’s like, we’re like, “shiny object syndrome.”
REID:
Yes.
BRET TAYLOR:
Some of them I am less excited about. And this one, I think even with the benefit of hindsight decades from now, I think will warrant the level of excitement that we feel right now. I think that the emergence of large language models has made artificial intelligence—which is something that we’ve been talking about for decades—reach a level of utility and practical application in so many domains simultaneously that I think it really will transform the way we use software. My daughter is almost 15 and a half, which is the age in California when you get your learner’s permit. And I am convinced that if it were five years from now, I would be chatting with an AI agent from my insurance company to add her to our insurance premium. Not tapping around a website.
BRET TAYLOR:
I am convinced that students around the world, whether or not they have access to great schools, will be chatting with an AI agent to help tutor them in any domain that they’re interested in. Even if it’s extremely esoteric or extremely complex. I’m convinced that, you know, the next time someone gets a lease agreement from an apartment, they can put it into ChatGPT and get great legal advice for free in a situation that no one would actually spend money on hiring an attorney for.
REID:
And by the way, they can do that today.
BRET TAYLOR:
All, all, everything I described—.
REID:
Yes, yes.
BRET TAYLOR:
You could do today. You know, I, so as I look at these trends, I am incredibly optimistic about the future because of artificial intelligence. I think it’ll have an incredibly democratizing effect on education, on access to medical advice, on access to expertise, broadly speaking.
BRET TAYLOR:
And as it relates to Silicon Valley—and, and I’ll try not, I’ll just pause after this, so it’s only 20 minutes—but I, I do think that the wave of excitement in Silicon Valley right now feels like the .com bubble. But you and I have talked extensively about this. I think the .com bubble was somewhat warranted. If you look at the top companies in the stock market today, how many of them were founded from 1995 to 2000? Amazon and Google alone represent a huge percentage of the stock market. If you look at all the insanity about how the information super highway and how it’s going to transform.
REID:
Pipes, pipes, pipes.
BRET TAYLOR:
It was basically all true. Right? It did transform commerce.
REID:
Yes.
BRET TAYLOR:
You know, maybe pets.com wasn’t a great investment, but Amazon was. Yes. And you look at Webvan and now you look at Instacart, DoorDash—all these other things, you know—it, it didn’t necessarily mean each individual company was transformational, but did the internet transform our lives as deeply as even the most frothy statements about .com?
BRET TAYLOR:
I think, yeah. The answer’s yes. So I think it’s, I think it’s important not to be cynical in these situations and recognize that there will be a lot of snake oil. There will be excess in investment. Lots of people will lose money. Lots of people will make money, and that’s the game of Silicon Valley. But if you zoom out big picture, will this technology truly transform the way we live and work? I believe the answer’s yes, and I think it’s pretty exciting to be alive right now and for the next 10 years, and see where it takes us.
ARIA:
And so, I mean, in our office, we are breathlessly following the next AI innovation, it feels like every day on our Slack channel. There’s, “oh, did you see this? Did you see this?” You’re at the center of it. What is like the next AI breakthrough that you’re so excited about or waiting for? Whether it’s personally, professionally for Sierra?
BRET TAYLOR:
I, I actually think that we will see continuous improvements in the quality and effectiveness of AI for the foreseeable future. With the one caveat that the foreseeable future is like two years [laugh]. I, it is incredibly hard to predict the future right now, even for those of us in the middle of it. But broadly speaking, when I look at the inputs to what is driving the progress in large language models and the frontier models that I think have come to define AI right now, that there’s lots of—we can go into lots of different directions about the applications of AI. It is first algorithms, so research insights. The most meaningful that probably Mustafa talked about yesterday was the birth of the Transformers model, which was an architecture from Google that they created sort of this current way of excitement. But there’s, every month, every year there’s new breakthroughs like OpenAI using chain of thought and reinforcement around chains of thought, which is what powers the new o1 model.
BRET TAYLOR:
And I won’t list them all. But with every new breakthrough and, you know, a new algorithm, you can take existing data and existing compute and create something that’s more intelligent. So that’s number one—is the algorithms. Number two is data. Some people think we’re running up against the limits of the textual data that powers a lot of these models. But we have a wealth of video and image and audio data that I think we’ve really yet to capitalize on. Really smart researchers are focusing on things like simulation—which won’t necessarily impact all domains, but if you think about things like self-driving or things like that, simulation has incredible applications. And then there’s synthetic data generation, which is very nuanced, but I think there’s a lot of really interesting applications there. And then the third thing is compute.
BRET TAYLOR:
And if you’ve looked at Nvidia stock price, you can see the level in investment in compute. And, I, I think that even with existing algorithms, existing data, you can train in different ways longer. You can use inference in really interesting ways as things like o1 have shown. And you can apply more compute and get more intelligence. So what’s so exciting about that is that if you think about a scientific breakthrough where the obstacle to the next milestone is single threaded, that’s where you can really stall out because you might lack the insight or technology to overcome that. When you have three separate and almost independent areas that you can make progress—if, for example, we run up into an architectural barrier on algorithms, you still have data and compute, and you can actually make progress there while you’re figuring out the next architectural breakthrough.
BRET TAYLOR:
Or if, you know, for example, the, the investment in compute stalls out because of the stress around CapEx, which I would feel if I were in the situation as well. You know, you can take algorithmic breakthroughs and data breakthroughs to make progress. So the reason why I’m so optimistic over the foreseeable future, which is admittedly short for me.
ARIA:
Yep.
BRET TAYLOR:
Is that when you have three areas making progress independently and all the smartest people in the smartest capital investing in them, I think the likelihood that something stalls out is very low. Just because I think we’ll make progress on one, two, or even all three of those things and continue to see the impressive breakthroughs that we’re seeing almost on a weekly basis right now.
ARIA:
Awesome.
REID:
So let’s give, give folks who don’t know a quick presee on what you’re doing with Sierra, and then talk about what the revolution with voice means.
BRET TAYLOR:
Yeah, thank you. I’m so, so proud of what we’re doing at Sierra. So at Sierra, we help companies make customer facing AI agents. So does anyone in the audience have a Sonos speaker? Yeah, a few. Well, next time you have a technical issue with it, which hopefully never happens, you’ll chat with the Sonos AI, which will help you debug and figure out what’s going on and help you fix it. Or if you have a family member setting up a Sonos speaker for the first time and has trouble with wifi, like seemingly all my family members, [laugh], the Sonos AI will help you figure that out too. If you have serious—
REID:
Did you arrange for your voice with your family members or? [laugh]
BRET TAYLOR:
Thankfully. No. You know, if you have SiriusXM in your car and you want to upgrade or downgrade your plan or figure out how to, you know, listen to the NFL, you’ll chat with Harmony—which is an AI agent powered by our platform. We power for across retail subscription services, help with sort of, I’ll say that frontline customer service, but really all parts of the customer experience. You know, the reason why I think just last week—and, and thank you Reid for noticing our announcement last week—we announced that our agents will not only support chat, but also can answer the phone. I think this is a really big breakthrough in customer experience. For those of you who probably many of you haven’t run a call center before, but if you did, you would know one metric—which is your cost per contact, which is essentially the cost all in when someone calls you and you answer the phone.
BRET TAYLOR:
For most companies answering the phone cost between $10 and $20, which is really expensive. And it’s, it’s mainly the wages of the people answering the phone. And it, it really depends on the complexity of the case and things like that. And that’s been roughly true for a long time. You can optimize it down, but it’s just an expensive way to interact with customers. And that’s why for most companies, if you try to find their phone number, it’s pretty hard. [laugh] You know, when something costs $10, you need it to, you know, essentially average out to roughly $10 of value back to you, and there’s not that many customer interactions that are actually that valuable. So I think a lot of people think of AI as a mechanism to reduce costs, and certainly that’s the basis of, you know, productivity enhancing the economy. What I’m so excited about for AI, for customer experiences, if you can bring down the cost of a phone call from 10 to $20 to an order of magnitude or orders of magnitude less than that, how many more phone calls will you have with your customers? You know, will you go to someone’s website and they’ll just be like, “call us. Call us anytime.” And, and, you know, I think if, for those of you—
REID:
And talk with us about anything, almost [laugh].
BRET TAYLOR:
Have a long conversation.
REID:
Yes.
ARIA:
[Laugh].
BRET TAYLOR:
And, you know, we can laugh about it, but, you know, let’s just put it this way. I think we all have, you’re all geeks, right? So we’re all geeks here. This is, we’re not, but like, imagine your family member trying to figure out why their Sono speaker isn’t working. Is it the wifi? Is it the hardware? Is it the software? Imagine not having to worry about being patient in that conversation. Like that, you can have the longest conversation ever. [Laugh] That is the most humane and empathetic thing possible. And I, I am so excited for that future. And so I think that often when you think about technologies, we often extrapolate from our current, current experiences and say, you know, I’m going to go from talking to an IVR system—which is like, honestly the worst thing ever, “press four for this”—to talking to an AI, which is better. I’m actually excited for the second order effects, which is, I think actually having interactive, personalized conversations with customers, even for consumer brands with tens of millions or hundreds of millions of consumers will become commonplace. And I think that’s going to be delightful. And I think we really haven’t contemplated the implications of that. And broadly speaking, I think every, for every, particularly those of you who run consumer companies, your AI agent will be just as important as your website in the future in terms of how you actually engage with your customers.
REID:
Actually more, but you, you know, that’s what you’re saying.
BRET TAYLOR:
Yes.
ARIA:
I love how the main tagline is like, now your parents won’t call you to set up your wifi. [Laugh]. It’s like a really positive effect.
BRET TAYLOR:
Look, I’m speaking to the audience, [laugh].
ARIA:
Yes.
BRET TAYLOR:
I have… in my career, I go home for Thanksgiving. Job number one is fix the wifi.
ARIA:
Absolutely. Absolutely. Well, so to your point, this audience, we have a diverse group. We have leaders of NGOs, we have folks who are doing AI fashion companies, we have legacy businesses, people who are in trucking. How should each of these people in our audience think about customer service, including a chat agent in their business? Like, what are the first steps? 15 years ago, like you said, “oh my gosh, we need a mobile app strategy.” How should they think about this next wave?
BRET TAYLOR:
I like to start with customer experience more than customer service. We work with a large number of furniture retailers. And for those of you who have recently gotten furniture delivered to your house, you probably ended up on the phone for 20 minutes and you’re trying to read your address and you’re like, “no, two words. Spring Brook Road.” You know, it is a, it’s a, it’s a miserable experience. And, and I think starting with all the moments that matter in your business and thinking, could we have something that was always on 24 7 that could instantly solve those problems and start there? You, you know, rather than starting with the highest volume cases, start with the cases that just aren’t delightful. And, you know, I think, you know, it’s interesting, most of our AI agents, actually, the customer satisfaction score and NPS score of our AI agents exceeds that of the contact center.
BRET TAYLOR:
And it’s not because of people in the contact center aren’t doing a great job. It’s because by the time you talk to them, you’ve been on hold. So you’re starting, you know, already frustrated, right? And so the benefit of AI right now is it’s multilingual, it’s empathetic, with access to your systems it can take action on your behalf. And if someone is having a negative moment with your brand, you can solve it instantly. And for those of you who’ve had moments like that where someone, a great associate solves your problem instantly—if you had something negative happening, it almost inverts your NPS score. You’re like, “wow, they solved my problem so quickly. I’m a loyal customer now.” So that’s really where I like to start, is pick the paper cuts in your customer experience and say, “if we have an AI agent that can actually instantly solve that problem, how can we turn those negative moments into positive moments?”
BRET TAYLOR:
And then broadly, I, I really like to just go through your customer journey, from pre-sales purchase consideration. All the way through a recent new customer, onboarding them into your, your product experience or service experience. To traditional customer service, something went wrong. Things like that. And just end to end saying, how can you, my co-founder Clay, likes to call it “paint by number.” You know, what’s next? You know, what’s the next thing that we can do to make something like this? And I think it’s really exciting. I mean, I think every company should have a customer facing agent. For those of you, do you know Chubbies, the shorts brand? Have you… You got to check out their agent. He’s named Duncan Smothers [laugh], and it’s, he is everything you’d hope Duncan Smothers would be. He’s got some personality.
BRET TAYLOR:
And so, I mean, these agents can, they’re, they’re fun, you know, like, this is a brand ambassador. This isn’t just, you know, providing action. So I think it’s incredibly fun. And, you know, it’s interesting, when I look at technologies, I love to think about the jobs that come out of it. For those of you as old as I am, if you remember in 1997, there’s a job called ‘webmaster’. You remember that? What a weird name, first of all, [laugh]. But now you fast forward, there’s web designers, there’s, you know, web architects, there’s all these jobs that came out of it. I actually think that conversation designers will be as important as UI designers in the future. You know, there will be folks in the customer experience side whose job it is to craft the behavior of these agents. A lot of our customers call that role AI architects, and they’re not technologists by trade. They’re on the customer experience side. So I’m just excited to see, sort of pull this thread in the future. And if most of your customer interactions are happening with a conversation with an AI, what are all the new roles you have at your company to make that an exceptional experience? So that’s super fun. We’re right in the middle of it.
REID:
And never thought I would have Duncan Smothers on the podcast, but, you know, it’s, it is one of those, those amazing discoveries. Let’s, let’s jump up a little bit since we’re, you know, kind of here live with a bunch of entrepreneurs at Masters of Scale. A question that I get frequently, and I suspect you do too, around AI and entrepreneurship is, look, is it all just big companies frontier? Or what is the role for entrepreneurs? And you exemplify as one of the instances of the fact that, you know, as I also tell lots of parties, journalists, other folks, that there’s tons of rule of, of room for AI entrepreneurship, even as the frontier models tend to be built at massive scale by a few companies. Say a little bit about that and what the opportunities for entrepreneurship around AI are.
BRET TAYLOR:
I’ll try not to be long-winded, but I’m going to start with how I view the market because I think there’s some nuance in what you said that’s important for entrepreneurship and where’s my place in this market? So I think the AI market will have echoes of the cloud market. So if you look at, at least my narrative around the cloud market is, you kind of had three specific domains. One was infrastructure as a service. So these are companies like Amazon Web Services, Microsoft Azure, Google Cloud, Oracle that are essentially investing ton of, a ton of CapEx to build data centers and then lease out that hardware to companies building software. And then you have makers of tools. So these are people selling pickaxes in the gold rush. Like if you’re moving to the cloud, what are the tools and technologies that you need?
BRET TAYLOR:
It’s tools like Snowflake or Databricks or Confluent, things like that. And then you have applications companies or solutions companies, software as a service. And obviously, you know, the one I’m most loyal to is Salesforce, where I spend a large part of my career, but there’s ServiceNow, Shopify, Adobe. There’s a, a really, really long tail of really specialized solutions for everything from procurement, to, you know, managing marketing spend. I think the same thing will play out roughly in AI, where we’ll have a very small number of companies building frontier and foundation models that will be companies that have the capital to have huge capital expenditures to invest in the compute to train these models. They will monetize it by—some will monetize it through consumer products, but I think a large number of them will monetize it by leasing it out to, to folks like, Sierra and all of you to take advantage of that technology.
BRET TAYLOR:
There’ll be a number of companies that provide tools. You know, if you’re building AI first software, what are the tools and technologies that you need for evaluation? Companies like Scale AI, you know, serve for really meaningful players in this space. And then there will be a very large number of people making solutions. And that’s companies like Sierra. Engineers on our team use Cursor as a coding copilot. There’s companies like Harvey, they’re creating legal solutions around AI. And I actually think that I am probably most excited and admittedly biased, but as it relates to entrepreneurship and applications and solutions. If you look at the past two decades of software, we’ve broadly moved from companies thinking “I should build software” to understanding that software is like a lawn, it needs to be tended to. And the moment a company builds software, they’ve essentially created debt that they will need to pay off for eternity.
BRET TAYLOR:
And as a consequence, most companies don’t want to build their own software. And most who have 10 years later replace it with a software as a service solution. And the reason for that is it’s just rational. You know, companies should focus on what the value they provide their customers, and I think it’s no coincidence that there’s such a meaningful long tail of really, I think, high scale companies—in this spirit of Masters of Scale—providing very specialized solutions to very specialized buyers within companies that just want to solve their problem. And I think that we’re just in the early days of AI. So, you know, two years ago, if you wanted to build something with large language models, all you could do is license a model directly. Now, and especially 2, 3, 4 years from now, I think most companies would provide—prefer to buy a solution to their problems, not buy technology. And I don’t think that fundamental truth will change. So I think there’s a lot of opportunity and tools. I think there’s relatively little opportunities and training foundation models, unless you’re one of the, the big players, and or one of the big research labs and partnerships with the big players. But I think there is infinite opportunity in building solutions to people’s problems that are facilitated by this amazing new technology.
ARIA:
Reid hit the nail on the head for one of the questions we get the most, but another question that we get from people on Possible is like, “ah, I love this amazing pro technology future. It’s so exciting. I’m an entrepreneur, but I’m also a parent and I’m so scared for my kids.” And the other night we were talking about what you would tell as a parent, what you would tell your kids to do differently. So you’re a parent, it’s this AI future, a lot of people in the audience are—what do you tell them about how to parent or what their kid should be doing or talking about? What’s that advice?
REID:
In addition to Duncan Smothers [laugh].
BRET TAYLOR:
So I get this question a lot and, you know, the questions are usually quite pointed, like, are software engineers going away? Should my child still study computer science? First of all, I just want to say I’m extremely optimistic about the future. At dinner two nights ago, we were, we’re having a great conversation, and I made the point that in 1776 when this country was founded the vast majority of people were farmers. And in fact, it’s a big part of the sort of founding story of this country, the abundance of land and farming and land ownership. You fast forward, you know, to now, and we’ve gone through two meaningful changes in our economy. First was the industrial revolution, where, you know, some things were automated. A lot of people worked in manufacturing, and now most economists would say we’re a services economy just by, you know, where the majority of Americans work. I, I would love to be able to like interview Ben Franklin and describe our economy and see his reactions. Like, like what do people do [laugh]? Like, I don’t get it. Like, how is working in a restaurant like sustain an economy? Like, I don’t get it.
REID:
And what is this podcasting?
BRET TAYLOR:
Yeah. Like, you’re a podcaster. Like I don’t get it. And, and I think it’s partly because we define our existence through the, like, the lens of our lived experience with the, the jobs that are available. And we’ve transformed many times as an economy. And, and it’s hard by the way. You know, it’s, if you look at the displacement in jobs and, and the rust belt, I mean, I don’t want to pretend it’s like transitions are easy and I think it would be completely insensitive to do so. But the idea that we’re giving up one thing and a new thing isn’t created, I just don’t truly believe. I think that we are incredibly creative and I think we will create new jobs that will be high, high leverage, high impact. And I’m really excited to be wowed 50 years from now by what our economy looks like and all the new jobs that will just be as foreign to me as perhaps podcasting would be to Benjamin Franklin.
ARIA:
Yep.
BRET TAYLOR:
Because I think it’s really exciting. So when I think about my kids—and I’ve got three kids—I, I come from the old school world where I think education is about learning how to think. And I think if there’s anything that’s important in age of rapid technology advancement, it’s not to become ossified in our tools. And what we do and how we do it does not define who we are. And I think the uncomfortable thing will be in our careers, I think will be an imperative that we re-skill and we learn new trades. We’re, an accountant and the spreadsheet was just invented [laugh], you know, and if you’re still on a slide rule four years from now, that’s not great. But if you’re the first in your office to learn how to use Excel, you’re going to be the most important person in the accounting department. And, and I, and I know that’s a trite example, but I think that’s going to impact all of our jobs. And I think, I think it’s exciting, and I think a lot of the entrepreneurs in this room, if you can create tools and technologies that enable people who aren’t experts to embrace these technologies, I think is an incredible opportunity.
REID:
And the other thing that’s great about this particular revolution is the AIs can help you with the re-skilling. As per agents, voice, and a bunch of other things and parental Thanksgiving visits. So with that, Bret, thank you very much. It’s awesome as always.
ARIA:
So that was the live show with Bret Taylor at Masters of Scale. And now we will get into the next part of the interview going both deeper and beyond the ideas we got into in the live show. So here it is.
REID:
So continuing our conversation from the Masters of Scale summit. You know, one of the things that we hope to kind of add into the full possible podcast is a little bit more of the kind of the depth of understanding of these tech trends that actually people who, you know, may not know you see, realize how many interesting vantage points you’ve had over your career. You’ve seen multiple technological inflection points, right? You know, Google, Facebook, Sierra and AI. What do these technological inflection points feel like? How should people kind of understand what comes from them? And then what is that understanding of the kind of these key kind of pivotal moments which transform industries? What does that mean for how you’re looking at AI now?
BRET TAYLOR:
It’s interesting that you bring up Google and Facebook because I actually would say from in my mind, they represent two slightly different approaches to creating companies. So in my mind, Google is a product of the invention of the internet and the web browser. And just like Amazon was and just like Salesforce was as well—where essentially the internet became widely distributed, the browser made an incredible product on top of the internet. And a number of use cases that I would say were intellectually obvious became apparent. Should you buy things via a web browser. Yeah, it makes a lot of sense. Right? Would it be useful to search overall the information on the internet? Yeah. Duh. That sounds great. Google wasn’t even the first search engine and you know, at the time there was AltaVista, AlltheWeb. There was a great company called Inktomi that powered search for a lot of the portals.
BRET TAYLOR:
It just turned out to be the best one, you know, thanks to page rank, thanks to a great business model. Similarly, Amazon wasn’t the only e-commerce site. There was buy.com, there’s half.com, there’s all these other sites. It turned out starting with books was a very, very smart strategy. And then the company just had incredible execution, you know, to, to become the best. In contrast, I would say Facebook—and a lot of the what have now been called Web 2.0 companies—largely were new product categories. You know, I don’t think of them, and I, and I might be wrong here, but I don’t think of them as being a byproduct of a technology shift necessarily to some degree. Once most of the world ended up on the internet, it created a market for connecting those people through LinkedIn or through Facebook or through others.
BRET TAYLOR:
You know, that probably wasn’t possible when there was fewer people on the internet earlier on, but in many ways it was really just a net new product category. In my mind, for that reason, when I think of AI, I actually look back at the .com bubble more than I look at that trend of Web 2.0 companies, because the way I would characterize AI market right now is large language models have created incredibly new technical capabilities that were impossible before. We can understand and summarize text at almost any volume, almost instantly. You can create conversational experiences and text and voice that actually feel as high quality or even higher quality than human conversations in some ways. You can use computers to do basic human level reasoning. When you combine all those things, some of the use cases are obvious [laugh], you know? Huh? You know, wouldn’t it be great that, you know, every product listing on the e-commerce side is personalized to my needs rather than being generic?
BRET TAYLOR:
Gosh, wouldn’t this be useful in the legal market? You know, where so much of that time is reading and comprehending text, which is something large language models are great at? You know, gosh, wouldn’t it be great to not have to wait on hold, you know, when I want to talk to customer service? And just like, buying things online was a novel idea, or searching the internet was a novel idea, a lot of it comes down to product and execution. I, that’s the way I view the AI market today. You know, to some degree the, the actual contours of the market and how it will play out, there will be some surprises. It would surprise me if there weren’t surprises, but a lot of the market is actually obvious. And the question is, what is the right product? What is the right go-to market commission? What is the company that is going to earn the right to be Amazon 10 years from now, or to be Google 10 years from now? And to me, it, it feels a lot like the .com market and less like the market where you’re creating new product categories necessarily. Because I think, simply put companies, in these categories will exist, and if one disappeared, another one would show up and take that market. Whereas I’m not sure that’s true many of the Web 2.0 companies that were fundamentally defining entirely new product categories.
REID:
Don’t you think that there will be new product categories out of AI—look like it’s kind of both with AI? There’s both like redefinition of a whole set of them because it’s kind of a, you know, a general purpose technology, ironically rhymes with GPT, to do all these things, but also the creation of new? Don’t, you, don’t you think it’ll be both?
BRET TAYLOR:
A hundred percent. And actually it’s interesting with the characterization I just gave, perhaps the Web 2.0, companies were actually, you know, second order consequences of the creation of the internet [laugh], you know, in some ways. And, and I think, you know, the, the analogy—and, and analogies have some flaws—but the analogy in the mobile era was, you know, the early phase of mobile companies were quite skew morphic and quite literal. And it took a few iterations of the app store to get more nuanced value propositions, like the ride sharing companies, like Instacart and DoorDash, and even, you know, companies like WhatsApp, you know, in, in, in many ways. So I, I think that there will be things that stand a reason as obvious market opportunities. And I think the things that will be most exciting are the completely new product categories.
BRET TAYLOR:
And I think the one that in circles that the three of us run in probably things like those personal AI agents are probably the most interesting category there, because they do fundamentally will probably change the way we interact with software. You know, where you have a piece of software that is an agent of you and you alone. And that’s very different than the way we interact with software today, which is fundamentally a tool. And, and I think that, so that’s the one that I am—I think a lot of people simplistically look at it like, “it will triage my email inbox.” I’m like that, it’s the, that’s the skeuomorphism version of a personal agent. I actually think it will be much more meaningful from that. And when you pull that thread more fully, it probably will change the way we interact with software in a really meaningful way. And I’m really excited to see those second and third order effects as well.
ARIA:
Well so, at Masters of Scale, we talked about your company, Sierra, and how it is revolutionizing the customer service industry and sort of turning it into this positive use case for all these companies. And I think a lot of people think that even though I would characterize us in like the first second of the first inning of AI, I definitely think there’s some people who are like, “well, we’re too late. Okay, well, the chat bots are out there. We’ve already done the AI agents. Like, where is there space for me?” And so you were just talking about all of these industries that you think are going to be transformed. You know, if you were giving advice or talking to entrepreneurs or just sort of speculating, what are the other areas that you think might really help grow this, you know, new AI universe?
BRET TAYLOR:
I think the early applications of technology tend to be quite literal and low level. So what I, when ChatGPT took the world by storm, including me by the way, a lot of the early startups that I saw were tools around that. You know, and very thin wrappers on top of the chat APIs made available by OpenAI and others. And similarly, I see a lot of startup that are selling proverbially, pickaxes in the gold rush. You know, like if you’re going to be building with AI, what are all the tools that you need? There’s fine tuning companies and there’s, you know, infrastructure companies leasing GPUs, and there’s, you know, data labeling companies and all these things. And the many of those will end up meaningful businesses. So I don’t mean to disparage them, but where I am most excited is people building solutions and applications that weren’t possible before, but are possible because of AI.
BRET TAYLOR:
And, you know, when we go back to the analog of the early iPhone apps, and you look at something like Uber, it was a combination of many things available. The GPS in your phone, the fact that it’s always in your pocket, the fact that it’s always connected to the internet—facilitated something that transformed transportation. Right now, I still see too many entrepreneurs looking at the technology and extrapolating from there. And I think that I am really excited for the wave of entrepreneurs who are looking at a problem in business or a problem in our consumer lives and solving it in ways that were technically impossible three years ago, and are possible now and transforming the way we work or transforming the way that we live. And I think those will end up some of the most meaningful companies here. Understanding nuance things like, I look at some of the great business model transformations like cost per click ads, you know, which I don’t think Google is the first, but certainly probably the most meaningful in terms of changing the business model of advertising or software as a service.
BRET TAYLOR:
You know, when Marc Benioff introduced that with Salesforce and, you know, against Siebel and others who are selling a perpetual license. I think entrepreneurs need to be looking at how do I solve real problems? How do I package those solutions in ways that customers actually will value more meaningfully, in that, you know, overused term in tech, but how do you actually create disruptive solutions with disruptive business models? And I do think we’re in the first second of the first inning. And I think the entrepreneurs who will, you know, transform the way we work and live will be the ones who are actually in with customers and focusing on problems that focused on technology.
REID:
And one of the things we talked about when we were on the Masters of Scale, you know, kind of summit stage was kind of the revolution that AI brings with voice. And obviously there’s some specific stuff—which we’ll have already covered—but I think going more deeply into kinda what is voice as the new front end for software, or what does voice mean for kind of revolutionizing our experience of the world and their mediation of the world or mediation of things through various digital contexts? Say a little bit about that.
BRET TAYLOR:
When I think about the impact of conversational AI and voice and how we use computers, I like to rewind and just look at the history of computers. You know, the first computers, you know, for, you know, were machines with wires that you connected, which quickly became punch cards, which were, you know, very close to that hardware. Which eventually, you know, became terminals where you could type the assembly language and eventually, you know, higher order programming languages to instruct them what to do. But, you know, you went from punch cards to terminals and then, you know, in Xerox PARC we pioneered the graphical user interface and you ended up with a mouse and a keyboard, not just a terminal. And I think the, you know, probably in my lifetime, the most significant change to the computing paradigm was the invention of the smartphone. And we went from sort of the era of mice and keyboards to the era of touch screens.
BRET TAYLOR:
And I think it was a really big change. It’s hard, you know, I don’t know if people measure what percentage of human computer interactions happen through a touchscreen versus a mice and keyboard, but it’s 90%, my guess, maybe higher. And it didn’t mean that mice and keyboard went away, but mice and keyboard became like a more like of a utility, a professional tool. And, you know, we send most of our emails on this thing and, you know, look at photos on this thing because it’s always in our pocket and always accessible. When I think about the impact of voice and what we spoke about before, the premise is that you can now speak to a computer and it will more or less understand what you’re saying and could behave correctly. Similarly, computers can converse in humane and empathetic ways. So for as a person, it’s actually quite delightful to talk to a computer.
BRET TAYLOR:
And if you look at the history of human computer interaction, we’ve tended towards the lowest friction, most convenient way to interact with software. So I am operating on the premise that if you fast forward five or 10 years, most human computer interaction will be conversational. And it’s not because it’s always the best solution. Just like typing an email with thumbs isn’t easier than typing on a keyboard, but it’s easier when you’re not at your keyboard [laugh]. And there’s so many opportunities to speak, you know, and I look at the number of microphones around us. I look at, you know, the prevalence of WhatsApp around the world and you know, how easy it will be for brands to publish their consumer experiences through platforms like that. I look at that wave of smart speakers we had a decade ago, and, you know, wonder if that will become increasingly relevant again, you know, as the smart speakers go from devices to get the weather to maybe devices that are like fully capable computers.
BRET TAYLOR:
And I’m really excited by that. And to the point we made on first and second order effects, I don’t think we’ve fully contemplated the implications of the human computer action being so seamless. I’ll say a couple things I find particularly interesting about conversational interfaces. If you just imagine an e-commerce website, with enough time, you could probably click around every page on the site. Go to every single shoe and shirt and pair of pants. And you could enumerate the finite list of things you could do and as user interface designer, literally what you’re doing is making a navigation to enable you to reach that finite list of items. You could draw a graph like a tree of all the places you could go. Human language is unbounded. So in this age of conversational experiences, you’re, you as a consumer have a lot more influence and a lot more agency into what those experiences are.
BRET TAYLOR:
And so I think we’re going to go from designing digital products that are finite and, you know, on rails to actually having, like listening to what your customers want to talk about and say, “wow, we should probably find a better way of answering those questions or fulfilling those needs.” And it’ll be a lot more interactive. Similarly I think designing experiences that have to account for all of conversational experiences, it’s an infinite landscape. And as a consequence, designing it, you have to give your software more agency. It’s impossible to enumerate everything your customers will say to you. So as a consequence, you actually need to define software in a much more flexible way, which means brand owners actually have to give up some control, which is kind of interesting too. I find it fascinating and I do wonder what the roles of people making digital experiences will be in the future. Conversation designers, AI architects, I believe we are creating an entirely new class of software and they will become completely dominant in the way we interact digitally.
REID:
Yeah, I couldn’t agree more strongly with what you just said there. I think it’s part of the kind of transformation of, of the kind of experience of the world. And you know, some of this I think will be in this kind of soft affordances. Like how much does the AI agent predict that when you interact, what you might be looking for, and then quickly adjust depending on it or, or prompt you like, you know, kind of, you know, it’s obviously one of the things that, you know, search has thought about for a while is, is could you show up and could I just tell you what you’re going to look for before you even told me that you wanted to look for it given contextual clues and, and other kinds of things as ways of doing it. And I think all of that will be part of this kind of conversational UI.
REID:
And I, I kinda liken it more to like, as opposed to, you know, it’s a tool and it’s a hammer and this is how you hit. It’s more like sharing a jazz improv or sharing a dance or a waltz and you’re kind of directing and learning as you’re doing it because you’re partially like, “oh, well, there’s a set of things that are about what I want and I want to do it, but I’ve, I’ve got this kind of agentic extension in dance.” And what that means for every single, like, if you add intelligence to every single thing that could possibly have intelligence, you, you mentioned speaker, but obviously lights and cars and all the rest of this stuff is going to be, you know, just kind of stunning. I think that kind of, that new paradigm of what is the world we live in and we kind of shorthand it to an agentic world and agentic interface.
REID:
For example, like here’s, here, here would be one prompt—because I’m curious to have you riff even more—but like, but people will kind of think, “okay, I’ll have an agent.” And you’re like, yeah, yeah, we’re going to have lots of agents, you’re going to have an agent for me, you’re going to have an agent for me at work. You’re going to have a work agent, you’re going to have a, you know, you’re going to have a CR customer service agent, and there’s, there’s going to be this, this multiplicity of agents and you know, that itself will be an entirely different kind of experience of the world. And so we don’t know, but, but if you were to, to cast the dart and say, you know, or a light of lens, you know, through the fog, you know, what, what would you say is like, well, here’s some things, here’s some good questions, here’s some things I’ve been thinking about in this new world.
BRET TAYLOR:
Yeah. Have your agent talk to my agent.
REID:
Yes, exactly. Yes. My agent will get back to you on that one.
BRET TAYLOR:
Yeah, it’s very hard to predict the future right now is the short of it. I think that I’ll give you my answer. And I also am looking forward to being surprised and I’m looking forward to being wrong because I find one of the most delightful parts of working in technology is all the unexpected outcomes. Because for me, it demonstrates just the creativity of entrepreneurs and the lack of, of imagination I think we often have with the second order effects of these technologies. So first, I hope in the future we spend less time looking at screens. As much as I have spent my career investing in technology and have benefited from the growth of the PC and the growth of the smartphone and how much it’s digitized, you know, the economy in ways that have been remarkable and wonderful. I do think right now we’re at this point where whether you have a teenager at home or you’re looking across at your spouse and you’re supposed to be having dinner, and instead you’re staring down at an email receipt that just arrived, we’ve created this incredible connected world.
BRET TAYLOR:
But we’ve let these devices kind of control us rather than, you know, having them really amplify us as human beings connecting with each other in some ways. And I don’t mean that in a pessimistic way. I think we all suffer from it now because we’re all hyper connected at all times. I, I do have a hope that, you know, when you have an agent that can actually help you deal with all of the influx of information available. And similarly, you know, you can interact with a computer not by having to touch a touch screen, but you know, perhaps by, you know, through a gesture or, you know, speaking through a, an AirPod in your ear—are there, you know, real transformational ways where that phone doesn’t end up in your hands so often. And you’re actually, you know, making decisions and delegating work.
BRET TAYLOR:
But the software that we use melts away a bit more from our lives. And I would be, I’m very hopeful that that will be the case just because there’s sometimes when you’re scrolling through TikTok because you’re bored and that’s perfectly fine. There’s other times when you’re, you know, getting a bunch of notifications because stuff’s happening at work and that’s the type of thing I think you can really imagine getting dramatically better. The other thing I’m quite excited about is democratizing access to expertise. We’ve all this experience where you’re like citing an apartment lease or something and you’re like, I’m not going to read that document, that legal document. It’s like, it’s probably fine, says there’s an apartment I have to pay for it. Probably some other stuff. It’s just below the threshold that any sane person would hire an attorney, right? Because you’d pay more for the attorney than it would be worth.
BRET TAYLOR:
But in those moments, if you can just say, “Hey, take a look at this lease for me, anything I should be concerned about here?” You get a doctor, you know, says, “Hey, here’s what I saw in your test results and here’s what I’m prescribing.” You know, if you could afford it, how would you feel about getting a second opinion on that? You know, you know, or even recommend where I could get a second opinion. Similarly, you know, you get the, I always like the, you know, the, you go get an oil change. Like you really need to replace your air filter. Like, do I, I don’t know. I, I don’t. And I, you know, and I think there’s these moments where, you know, we can’t be experts in everything, especially in an increasingly complex world. And having these agents that can actually have that expertise and democratize access to it, I really think it’ll be empowering to individuals and I think it’ll be empowering to individuals in a way that really crosses sort of access to sort of economic lines as well. Because this technology is sort of naturally democratized.
REID:
I’ve, again, strong plus one, everything you just said. You know, one of the things that I’ve been thinking about in this kind of what are the things that I think are almost certain that seem very weird and a little alien today? Like, what is that future human experience that’s totally natural? It’s like, for example, you know, the smartphone thing is like, if you had told people, like, it’s like everyone’s going to be kind of on their smartphone, you’re like, and it’s going to be interrupting, like even family dinner and so forth. People are like, “well, that sounds really alien.” And it’s like, okay, it has some sub-optimum things. You have to learn the discipline of not looking at your phone and, and, you know, that kind of stuff. But, but ultimately, very good. I think a similar one to that is I think that we are going to, within a small end number of years, very quickly get used to one or both of us basically has our agent on in our conversation, right?
REID:
Like, because it’ll just be like taking notes and remembering and going, “oh, we should do this follow up,” but not, not, work context for sure. But also it’s kind of like I’m kind of following along in case something occurs that’s kind of important that I then remember or note or I go, “oh, hey, remember, you know, when you were talking to Bret about this, this might be something to follow up on.” Or, or, “oh, Reid you just said something really dumb. It’s actually not true, right? I went and looked and, and here’s this thing.” And I actually think that we’re going to, we’re going to not just move from, “oh my God, that is weird and alien. Like how, how in the world could we possibly want that as like always being surveilled or monitored,” to, “Hey, that’s really useful.” And as long as it’s, I get the conviction that’s on my side, it’s, it’s for me, it’s with me as opposed to on me. Then I think, I think we’ll see that behavior adopted a lot in a not large number of years.
BRET TAYLOR:
Yeah. A corollary or maybe just an orthogonal point to that is one of my theories is as these models become more powerful, it will increase trust in software. These models that are more powerful are more capable of not misbehaving, like hallucinating or being susceptible to things like prompt injection attacks. They’re also more capable of having comprehending very complex and nuanced problems. And you know, I believe that, you know, as these models become more powerful, more people will trust them, not less because their power will actually engender trust in the people using them. So I agree with you. I think AI will be integrated into our personal professional lives in a really deep way. And I think people really will trust them and, and rightfully in my mind, because I think these will be architected in a way that are preserve the goals of the person using it, whether it’s, you know, privacy and security or, you know, whether it’s—it’ll, it’ll have you in mind. And I think I’m very optimistic about that because I think the more I’ve encountered more powerful generative models, the more I would trust it with agency.
ARIA:
I think it’s interesting because most people, if you—well, maybe I’m wrong—but I think if you asked many a person on the street, they’d be like, “well, AI is going to decrease trust. AI is going to give people less agency. AI is going to flatten the internet. So everything’s the same.” And we’re saying a lot of the opposite. Like, I love the idea that like, instead of calling that like competent friend, you have to be like, “Hey, have you ever read a legal agreement before?” Like everyone will have their competent friend, you know, in the form of their agent. Or, you know, I would argue that the internet is being flattened right now. Everyone I know just uses Wirecutter to buy their bathroom mat, to buy their iron, to buy their refrigerator. Like, we’ll actually have so much more uniqueness and creativity when we’ll have an agent that knows me and knows, “no, I don’t want the Wirecutter pick. Like, I always want something, you know, X, Y, Z.” And so Bret, you are thinking about AI all the time. For those people who are listening who aren’t thinking about it, like what should they be thinking about that they’re not? Like, what do you often hear in conversation that you’re like, “huh, I think about that differently. Or, here’s something I think about all the time that’s not getting covered.” Like, what insights can we learn from you who’s living and breathing this stuff?
BRET TAYLOR:
Well, first, I think when you create dramatically powerful new technologies, some things that used to become, well used to be expensive, become inexpensive, and some things that used to be exclusive become more universal. You know, just let’s just take the smartphone. When the smartphone was created in 2007, there was on the order of 2 billion PCs in the world I believe. And in the western world, many people had personal computers, but most people who had access to computers at all did so in a shared way. Internet cafes, things like that. We now invented these technologies where everyone has a supercomputer in their pocket that’s connected to the internet, and it democratized access to the internet, to information. It created digital payments infrastructure that have really changed, particularly in the developing world. You know, there’s sort of leapfrog many parts of sort of the traditional evolution of the financial system, things like that.
BRET TAYLOR:
What’s also interesting about the smartphone is no matter how much money you have, everyone’s smartphone is basically the same [laugh]. There’s not like, you know, some tech trillionaire—their smartphone’s the same as yours. You know, it’s just like there’s—so, it’s really interesting. You went from something that was like, very, very exclusive before to something that’s almost universal at this point. I think there’s more smartphones than people in the world at this point or darn near close to it. And there’s absolutely zero differentiation between everyone’s smartphone. Like it’s just—and, and that’s really interesting when you think about how quickly that changed. I think when I think about AI, I think about through a similar lens, which is it’s taking things that were, you know, maybe really hard to do and making them commoditized, which in the short term might disrupt a job that’s related to that exclusive process.
BRET TAYLOR:
It might change the business model of a company. It might eliminate the business model of a company that was maybe providing access to that exclusive service that’s now easy. But just like the adoption of digital technologies, it, it created a lot of new industries and jobs as well. And I don’t think, you know, 20 years ago you knew what an influencer was or that, that you could make money off of it. Let alone more nuanced things like I’m an SEO expert, which is actually a really valuable job to do right now. And it was created because of this industry that was created around it. And so in conversations I have people often look at the current state of our economy and our jobs and the way we work, and think about what might be removed from that, but don’t think enough about what will be added to it.
BRET TAYLOR:
And the other thing, and the reason why I have so much optimism is, I think if AI will drive productivity in the economy in a really material way, it will lower the costs of a lot of services in a way that I think will really benefit humanity. You know, I think it—healthcare costs remain very high and no matter which side of the political spectrum you’re on, whether forget whether it’s nationalized, my point is that we can reduce the cost of healthcare. More people will have access to it. And any construct of, of the healthcare, you know, industry, you know, reducing access, the cost of access to high quality information, mental healthcare, legal advice, whatever it may be, will really benefit humanity. And so I think, candidly, I think people are maybe too focused on the really, really short-term implications and not thinking enough about how much this will really improve the lives of millions and millions of people around the world.
BRET TAYLOR:
And all that said, and, and we talked about it a little bit on stage, it doesn’t mean that there won’t be short-term disruption that won’t be uncomfortable for many of us individually. And that’s, that is a, a real issue. But I also think that, you know, we will hopefully look back on this moment as the birth of a technology that really improved the way we live, our longevity, our mental health, and, and it will take a lot of work to ensure we can mitigate the downside risk as well. But I’m really, really optimistic about it.
REID:
Well, all this thread very naturally leads us to our last question before the rapid fire, which is, you know, kind of when millions people are getting access to new tools that dramatically increase the options and opportunities, which we have been, you know, covering in some breadth and depth here, we call that superagency—because you, you benefit from both your own new capabilities and the new capabilities of others. So what’s one aspect of superagency you would want to play out in a compelling way?
BRET TAYLOR:
One thing I’m extremely excited about is an unknown artist or thinker having a dramatically huge impact on the world. You know, a 16-year-old creating a film of the scale of Interstellar, the equivalent of a Ramanujan, who thanks to access to, you know, their agent, were able to make a breakthrough discovery that previously would not have been done from a person who otherwise, without the existence of these tools, would not have had access to the, the tools or information to do so. Because I think that will be the moment when I think a lot of society realizes just how empowering these technologies are. You don’t need a VFX studio to blow up that spaceship in space. You just need to describe it in a prompt. You don’t need to have access to a professor at Cambridge to, you know, explore higher math. And, and I think that, that that’s what I’m most excited about, and that individual, those individuals will come to represent, I think, the opportunity afforded by this technology.
REID:
That’s a very natural transition to our first rapid fire question, which is, is there a movie, song or book that fills you with optimism for the future?
BRET TAYLOR:
This is so funny because you said optimism, and this is one of the most grueling books I’ve ever read. But there’s a book called Endurance about Shackleton’s quest to the South Pole. But I think about it a lot in the context of starting a company and building here because it demonstrates like the resolve a group of people can have in it. And so, I think in part because the, I’m in the middle of starting a company right now, I think a lot about that. And I find it quite inspiring despite all the seal meat, [laugh] about 200 pages in the middle of that book. But it’s just a fantastic book. And I think probably the greatest story of just like, you know, having an adventurous mindset and truly having endurance,
ARIA:
True endurance. So next rapid fire. What is a question that you wish people would ask you more often?
BRET TAYLOR:
To our previous conversation, what do you think the most surprising new type of career will be in 10 years? I don’t have an answer, but actually one of the things that Reid and I have, eat a lot of meals together and have conversations, and it’s, I think it’s a perfect Jeffersonian question. You know, where by the time you get to around a table of smart people, you probably would’ve changed your worldview. So I find questions that I don’t know the answer to more interesting than questions I do know the answer to.
REID:
Amen. Where do you see progress or momentum outside of your industry that inspires you?
BRET TAYLOR:
I’m really excited about progress in battery development. I, we just put in Powerwall at our house and solar panels. We’re, I drive an electric vehicle and I, it’s been really great to see the world have some breakthroughs here. And as I think about the development of sustainable energy, just how important batteries are to that future and, and energy storage. So I know nothing about it. I, you know, I read books about it just out of curiosity, but I’m really excited to see just how much rapid progress there’s been and how that changes the way we think about energy in the future.
ARIA:
Couldn’t agree more. If energy becomes cheaper, to your point about AI, what are all of the amazing things and amazing business models that we can see in human innovation? So plus one to that. So always our final question. Can you leave us with a final thought on what you think is possible to achieve, if everything breaks humanity’s way, in the next 15 years? And what’s our first step to set off in that direction?
BRET TAYLOR:
I think if everything breaks our way, every child in the world will have access to exceptional education. And I think it’ll really change the access to opportunity in a way that previously was almost an intractable problem, particularly at, let alone a global level, even a state level or, or national level. And I think the opportunity to provide access to, to education is, I think one of the most amazing things we can do as a society, because I think, you know, especially here in the United States, we’ve sort of found it on the premise of equal access to opportunities and, and I think it will probably do more than anything else to provide that.
ARIA:
Thank you, Bret.
REID:
Bret, as always a great pleasure. Thank you.
BRET TAYLOR:
Thank you.
REID:
Possible is produced by Wonder Media Network. It’s hosted by Aria Finger and me, Reid Hoffman. Our showrunner is Shaun Young. Possible is produced by Katie Sanders, Edie Allard, Sara Schleede, Vanessa Handy, Alyia Yates, and Paloma Moreno Jiménez. Jenny Kaplan is our executive producer and editor.
ARIA:
Special thanks to Surya Yalamanchili, Saida Sapieva, Ian Alas, Greg Beato, Parth Patil, and Ben Relles. And a big thanks to Ali Aills and Little Monster Media Company.