No Priors: Artificial Intelligence | Technology | Startups - Going Full Send on AI, and the (Positive) Impact of AI on Jobs, with Kevin Scott, CTO of Microsoft

Episode Date: May 24, 2023

In this episode, Sarah and Elad speak with Microsoft CTO Kevin Scott about his unlikely journey from rural Virginia to becoming the driving force behind Microsoft's AI strategy.  Sarah and Elad discuss the partnership that Kevin helped forge between Microsoft and OpenAI and explore the vision both companies have for the future of AI. They also discuss yesterday’s announcement of “copilots” across the Microsoft product suite, Microsoft’s GPU computing budget, the potential impact of open source AI models in the tech industry, the future of AI in relation to jobs, why Kevin is bullish on creative and physical work, and predictions for progress in AI this year. No Priors is now on YouTube! Subscribe to the channel on YouTube and like this episode. Show Links: May 23, 2023: The Verge - Microsoft CTO Kevin Scott Thinks Sydney Might Make a Comeback May 23, 2023: Microsoft Outlines Framework For Building AI Apps and Copilots January 10, 2023: A Conversation with Kevin Scott: What’s Next In AI Sign up for new podcasts every week. Email feedback to [e-mail] Follow us on Twitter: @NoPriorsPod | @Saranormous | @EladGil | @kevin_scott Show Notes: [00:00] - Kevin Scott's Journey to Microsoft CTO [12:44] - Microsoft and Open AI Partnership [21:18] - The Future of Open Source AI [32:12] - AI for Everyone [45:29] - AI and the Future of Jobs [51:44] - The Future of AI and Regulation [58:10] - Taking a Global Perspective

Microsoft, the BMF Productivity Cloud and Gaming Company, has taken a massive bet on AI. Everyone's paying close attention to its partnership with OpenAI, and the technical community has been amazed by its release of some of the first truly useful and broadly deployed AI products such as GitHub co-pilot. Its full-on attack on web search with the new LLM-powered BingChat is making its incumbent competitors dance. Today on No Pryors, we're thrilled to speak with Kevin Scott, CTO of Microsoft, and the driving force behind their AI strategy. Kevin's leadership, both at Microsoft and Pryor at LinkedIn, Google, and AdMob, as a technologist, is especially inspiring to me, given his distance traveled from his childhood home in Ruegel, Central Virginia. In 2020, he published a book, reprogramming the American Dream about making AI service all. Kevin, welcome to No Priors.

Thanks so much for joining us. Thanks for having me, guys. Can you start by sharing with us some of your story? How does one go from a farming community in Virginia where your parents didn't attend college to the CTO of Microsoft? I don't know. I think it is a very unlikely journey. It's like certainly not a thing that I ever could have imagined. I think part of it is I was just super fortunate to be wired like a nerd and growing up when I grew up.

So, you know, when I was a teenager in the early. early 80s, personal computing was happening, and, like, that was the thing that I happened to fixate on. And even though we were relatively poor, I managed to, you know, scrape together enough bucks to get myself a personal computer that I could have and just tinker with all the time. And it was, like, it was a Radio Shack Color Computer, too, like one of these things with chicklet keys that you actually connected to a television. Like, I had it hooked up to a 13-inch TV, and it had a cassette recorder that you stored and loaded your programs on. And it was just the thing that I was obsessed with, and I stayed obsessed with computers from then on.

And it was just me trying to find a path at each step where I could work on the most interesting thing that someone was dumb enough to give me permission to go work on. And again, it's a lot of luck. Like, there's no way you can plan a path from rural central Virginia to CTO of Microsoft. But, you know, I think it does help to have a high-level vision in your head for what it is that you want to do. Like, just knowing what you're aiming for always helps. What was that vision for you besides, like, you know, obsessed with computers, wanted to work on them? Yeah, I more or less had two of them. So the first vision I had, when I was a teenager, was I wanted to be a computer science professor.

So I just looked at what computer scientists did and thought this is the most amazing stuff I've ever seen. And I went to a science and technology high school. And the way that it worked where I lived is like a really rural area. And so the science and technology was a governor school. So it was centrally located in each high school. in these four or five counties that surrounded the governor's school, got to send two students each. And so I was one of the two students that got selected from my high school to go to this thing. And my computer science professor there was this guy, Dr. Tom Morgan.

And I just sort of felt like he'd opened up this entire new world to me. Like it was just thrilling to learn all of this stuff. And I was like, yeah, I want to be like Dr. Morgan. And a lot of this stuff for me is about, you know, like who those influential role models have been in your life. And so as soon as I like met Dr. Morgan, I was like, oh, I should just go be a computer science professor. And that was the path I was on until I was about 30 years old when I, you know, I was a compiler optimization and computer architecture programming languages person. And I got pretty disillusion with what being a computer science professor actually was, to what I wanted to do.

I just wanted to have a lot of impact. And my perception at the time when I was making these decisions was that you can have a lot of impact as a computer science professor. And the impact was actually great, but it wasn't the impact that the system appreciated. So the impact that you can actually have is inspire students to go pursue these careers and they will go on to do much greater things than you've done yourself.

And like that, that to me was the greatest impact, but it was the least appreciated part of being a computer science professor, like back in the 2000s when I was making these big decisions. And so I decided to leave. And I didn't at the time know what next actually was going to be. Like it had been my mission for almost 15 years at that point. And like I was a little bit lost. and I saw that a bunch of my academic buddies were all working at this startup called Google, and I didn't understand why they were working at Google. Like, Google was, like, some little box, and you typed keywords in, and it gave you 10 things, like, how is that hard? But, you know, it was Holtzla, who was a compiler person,

and Jeff Dean, who was a compiler person, and Alan Eustace, who was a compiler person, like all of these people who, you know, who I went to conferences with and whose papers I read. And I was like, all right, well, maybe I should send my resume in. And, like, I sent my resume in and got called to do a bunch of interviews. And they, like, it was the best interviewing experience I've ever had because they took what must have been every compiler person in the company at the time and put them on my interview panel. And I was like, oh, my God, this is amazing.

Like, I had the best day interviewing there. And I got this job offer, and I went, I got this choice. They just started Google New York, which was the first office outside of Mountain View. And they were like, you can come to Mountain View or you can go be, you know, the 10th person in this New York office. And my wife and I wanted to live in New York more than we wanted to live in Mountain View. And so that's what we did. And after I got there, this is where the new mission came in. I just, so we were hiring these brilliant, brilliant.

people at the time. And the way that we did hiring was kind of crazy. It's like, all right, well, if you're smart, just come work here. And like, we have no idea, like, what exactly it is you're going to do. And you, like, came in and you sort of sorted yourself out. And we had these people who were so accomplished and so brilliant. And they would come in and choose to work on things that just were going to have no impact at all. Like, they were intellectually very interesting, but they were just sort of silly in that they were never going to connect with anything that moved the needle for the company, which was exactly the problem I was trying to get away from, you know, in being like a research computer scientist. And so I sorted myself

out, like I found a pragmatic thing to go work on. Like, you know, I won't go into the details of what it is. But, you know, like the whole team won a Google Founders Award, which was a big deal for like solving this like very sort of unsexy problem with a bunch of very fancy computer science, which was one of the things I think Google, Google did really well. And then I was like, okay, well, I should just go help more people sort themselves out as well. And that's when I became a manager. And then from that point on, it was all about like, hey, I want to, I want to help as many engineers as I possibly can, like, make sure that their work lines up with something that's,

you know, both interesting and meaningful. I think that it's actually pretty under-discussed a degree to which early Google had so many academics actually running important parts of the company. Yeah. I think ERS is a great example, and I think there's others. And so I haven't actually seen anything like that since until maybe now, more recently at Open AI, there's more academics. So, you know, you feel like the research community is popping back up again. But it's been maybe a decade or too since that's happened. Yeah, I mean, I think that that's actually a really, really great observation. So when I go sit in OpenAI, it really reminds me of early Google days. And it's about the same size Google was when I joined. And so like I couldn't figure it out

for a while. And I was like, wow, this is like really giving me like, you know, early Google nostalgia. And, you know, the conclusion to draw from that is like not that they're the same companies are they trying to solve the same problem? It's just sort of the energy of the place and like who they've chosen to hire and like. Yeah, it's the first time I've seen like string theorist getting hired again. Yeah. And the computer science roles. Yeah, 100%. You know, since Google days. Yeah. You and I like probably both worked with Janitin Zunger who works at Microsoft right now. Like I remember like it's like, all right, Jonathan's working on this big distributed file system stuff and like what's his degree? Oh yeah, he's a like string theory guy. Yeah. So

A big part of your mission for, you know, the last decades has been helping string theorists and other engineers figure out how to be useful in their orcs. The other part seems to be, of course, like actual technical direction, right, deciding like what's worth investing in. And you've worked on machine learning products for a really long time, like ads auctions at Google, recommendations at LinkedIn, et cetera, et cetera. Was there a moment when you decided or you realized personally that AI should be a key technical bet for Microsoft?

Yeah, I mean, I've been at Microsoft a little over six years now, so almost six and a half years. And pretty quickly, it was obvious that AI was going to be very, very, very important to the future of the company. I think Microsoft already understood that before I got there. And then it was just how do you focus all of the energy on the company on the right thing? Because we had a lot of AI investment and a lot of AI energy, and it was sort of very diffuse when I got there. So no lack of IQ and actually no lack of capital spending and everything else, but it was just kind of getting peanut buttered across a whole bunch of stuff. And so the thing that really catalyzed what we were doing is, I mean, maybe this is a little bit too technical, but like, you know, we, before I got there, the technical thing that had been happening with some of these AI systems that, to me was very interesting as transfer learning was starting to work.

So, like, you were going from this mode of, you know, the flavor of statistical machine learning that I cut my teeth on, like, in my first projects at Google, which was, you know, you have a particular domain of data and, like, you have a particular machine learning model architecture that you are, you know, your training and, like, a particular way that you're going to go do the deployment and measurement and whatnot. And it's all, like, you know, siloed to a like a use case or a domain or an application, to seeing AI systems that you could train on one set of data and use for things for multiple purposes. And you saw a little bit of that with some of the cool stuff that DeepMind was doing with reinforcement learning, with play transfer across some of the

gaming applications that they were building. But like the really exciting thing was when it started working for language with Elmo and then, you know, Bert. and Roberta and Turing and, you know, like a bunch of things that we were doing. And that was the point where there were so many language-based applications that you could imagine building on top of these things if it continued to get better and better. And so we were just sort of looking for evidence that it was going to continue to get better and better. And as soon as we found it, like we just started like all in.

That was everything from doing a partnership with OpenAI to, you know, like at one point, I seized the entire GPU budgets for the whole company, and I was like, we will no longer peanut butter these resources around. Like, we will focus them because it's all capital intensive. It's like we will just allocate these things to things where we have really, really strong evidence-based conviction that, like, a particular path is going to benefit from adding more capital scale. I remember it must have been like five years back now. We were at dinner, and now GPU capacity is the talk of the technical town, right? but you were, like, I asked you what your, like, most pressing issue was, and you're like, how am I going to spend on GPUs this year and how I'm going to distribute those

GPUs? Yeah. Yeah, and it was, and it has been. It certainly hasn't gotten any easier. But, I mean, so, Eli, like, I think, you know, the question you were asking is, like, how we decided to do the Open AI Partnership. And so, like, the reason that we did the partnership was twofold. So one is, with transfer learning actually working, you can imagine building a platform.

for all of this stuff, so that you're building single things where you're amortizing the cost of the things across a whole bunch of different applications. And because we have a hyperscale cloud, like one of the things that I was really, really interested in and, like, beyond interested, like it felt, you know, just like an existential thing is how do you make sure that the way that you're building your cloud all the way from, you know, your computing infrastructure, your networks, your software frameworks and whatnot, how can it really serve a whole bunch of interests beyond your own? And so, like, we felt like in addition to the high ambition things that we were doing inside

of the company that we needed, like, high ambition partners. And when we looked around, like, Open AI was clearly the highest ambition partner that was in the field, you know, and I think still their ambition is just breathtaking and what it is that they're trying to accomplish. And so that was one thing. And then the second thing was, like, you know, they really had a very similar vision to the one that I had about, like, these things were evolving into platforms. And, like, we were able to, because we were so aligned on vision for the future, like, we could figure out how to do a partnership where, like, even though, like, there's just a ton of difficult things. And, like, you know, I think there's probably some conservation law of, you know, the stress from difficulty.

So it's not like it ever goes away, but, like, it's sure. stress in service of a common goal. And like, that's the thing that make good partnerships work. I think one of the stunning things about the partnership in some sense was the timing, because if I remember correctly, Microsoft made its first investment or its first significant investment in OpenEI right after GPT2 launched, or right around GPT2. And this is before GPT3. And there was such a big step function between the two of them that I think it was less obvious in the GPT2 days that this was going to be as important as it was. And so I'm a little bit curious, like, what were the signs that made you decide that this was a good partnership to have versus building it

internally versus you know usually as a larger company there's the old like buy build partner kind of thinking and so i'm just sort of curious like how how you all decided to to partner in this moment of time where it's very non-obvious and you invested a large sum of money behind that yeah there and i like i don't want to like have revisionist history and like paint a rosier picture than there actually was so that there was a huge diversity of opinions inside of the company on the wisdom of doing doing this. And so Satya, like, has this thing that he talks about, like, no regrets investing. So, like, there are things where you do the investment and, like, there are multiple ways to win and, like, you even win a little bit when you lose. And so this was one of those no regrets

things in that, like, the very, very worst thing that could happen is we would go spend a bunch of capital on computing infrastructure. And we would learn. what to do at very high scale for building these AI training environments. And, you know, you'd have to believe something very strange about the world of AI that you wouldn't need advanced computing infrastructure. And then there were just multiple ways where, you know, like, and we had a bunch of evidence that, you know, we had gathered ourselves and that Open AI had that gave us, you know, which unfortunately I can't talk about, but like that gave us, you know, pretty reasonable

confidence that scale up was actually working. You probably seen the famous OpenAI compute scale paper where they sort of plot on the log scale, like how many petaflop days or whatever the unit of total compute they were using on that graph that shows from 2012 when we first figured out how to train models with GPUs through, I think the plot ends sometime in 2018. Yeah, that we're basically consuming 10 times compute. more compute every year for training state-of-the-art models. And so, like, you know, I just had super, super high confidence that we were never going to get to the point

where we're like, all right, we got enough compute. It was a very bold move. I think it's very striking all the amazing things that Microsoft has done over the last few years in terms of just incredibly smart strategic moves of the time didn't seem obvious. And now are just in hindsight, you know, really brilliant. I guess that more recent move

as you announced a collaboration with Nvidia to build a supercomputer part by Azure infrastructure combined with NVIDIA GPUs. Could you tell us a little bit more about your supercomputing efforts in general and then maybe a little bit more about those collaborations, both NVIDIA and Open AI on the supercomputing side? Yeah, so we built our, the first thing that we called an AI supercomputer. I think we started working on it in 2019, and we deployed it at the end of that year.

and it was the computing environment that GPT3 was trained on. And, yeah, we had been building a, like, a progressively more powerful set of these supercomputing environments. Like, we built them in a way where, like, the biggest environment is just because, you know, they're very capital-intensive things tend to get used for one purpose. But the designs of these systems, like, we can build smaller stamps of them, and they get used by lots of people. So we have tons of people who are training very big models

on Azure compute infrastructure, both folks inside the company and partners who can come in. And it was the thing that was not possible to do before, where you could sort of say, like, hey, I would like a compute grid of this size with this powerful network to do my thing on. And so, you know, and Vidi has been, you know, our compute and network partners since they bought Melanox, you know, for years now. And the thing that makes that work is generation over generation, like you're just getting better, you know, price performance from the systems.

And we work super closely with them, like, defining, you know, what the hardware requirements need to be, you know, in the coming generations of GPUs. because, like, we have a pretty clear sense of where models are going and, like, what model architectures are evolving towards. And so, yeah, I mean, it's just been a super good, super good partnership. Yeah, like, we're, we're deploying a hopper now at scale and, you know, like a bunch of the features of hopper, like, you know, 8-bit floating point, you know, arithmetic and a bunch of other things are, like, things that, you know, like we've been planning for for a while. Yeah, I guess one last question on sort of this, both supercomputers as well as platform side of things, is I'm a little bit curious how you view the world shifting in terms of close source and open source models and, you know, the mix that will exist because obviously from an Azure perspective, lots of people are running open source models on top of Azure right now. Yeah, I mean, it is an interesting thing that people are framing it as some kind of binary thing. Like, I think you're going to have a lot of both. Like, we still don't.

see any reason to believe that you're going to want to not build bigger models. But like we just know in our own deployments. Like if you look at things like BingChad or Microsoft 365 copilot or GitHub co-pilot, you end up using a portfolio of models to do the work. And like you use it for performance and cost optimization reasons. And you use it for, you know, just sort of precision and quality reasons sometimes. And so there's always this melange of things that you're doing, and it's never either or. I'm actually really excited by what's going on with the open source community.

I think my biggest question mark there is how you go deal with all of the REI and safety things. But if you look at the technical innovation inside of the open source community, it's really thrilling. And, like, we, you know, like we were doing some cool stuff right now. Like, I was just playing around yesterday with that $12 billion parameter, Dolly 2.0 model from Databricks, which, like, runs quite nicely on a single machine. And, like, yeah, I'm still enough of a dork to, like, love playing around with things that run on single machines. Like, it's, you know, really, really impressive work. Yeah, yeah, yeah, it's super cool.

How do you think about that from the context of enabling AI for your business customers outside of your core product? So is there a specific sort of B2B AI stack that's coming? Are there specific tools coming? To your point, there's safety, there's analytics, there's fine tuning. There's so much stuff that you could potentially provide. I'm just sort of curious how you think about that. Yeah, I mean, I don't want to turn this into some kind of weird marketing spiel. But, you know, we have this point of view that we started with this assumption that

AI is going to be a platform and the way that people are going to most usefully make, or the way that people are going to make most use of the platform is by building tools that assist people with jobs. So it's like less about these fully autonomous scenarios and more about assistive tech. And so the first thing that we built was GitHub co-pilot, which is a coding tool. It's a thing where you can sort of say in natural language what you would like a piece of code to do and it emits the code. And then you, you as the developer, like the same way that you would take a suggestion from a pair programmer, like you scrutinize it. code review it and, you know, decide whether or not it makes sense for your application.

And, you know, like, that was the first version of GitHub co-pilot that does a bunch of other things now. And so the thing that we have observed is this copilot pattern is actually pretty, you know, pretty generic. And we built a bunch of co-pilots since then. And the way that we built them, like there's a, there's a copilot stack that looks almost like one of these OSI, you know, networking diagrams, and it starts with a bunch of user interface patterns that you have, like they're now an emerging plugin ecosystem for, like, how you extend the, you know, the capabilities of a co-pilot for things that you can't natively get out of the model. And then it is a whole stack of things, you know, sort of an orchestration mechanism

like Lang Chain is one of the popular open-source orchestrators, but there are a bunch of open-source orchestrators. We have one that we've developed called Symantecernel, that we've also open-source. There is this whole fascinating world right now that didn't exist nine months ago around prompt construction and prompt engineering. And so, like, there's an entire art form

and a set of tools that people have access to to design a metaprompt, which is sort of the standing instructions to the model to, like, get it to conform itself to the application context that it's in. Like, you have these new things, like new software development patterns like retrieval augmented generation or rag is, like, you know, we were doing this before. It had a name on it. And, you know, so it's basically a way to, like, take the prompt that's flowing from the application and to inject context into the prompt that will help the model better respond. And then there's a whole bunch of safety apparatus that you have.

So it looks a lot like filtering on both the way down as the prompt flows through the stack all the way down to the model as well as it flows back up. So what things are you not going to let the application or the user send all the way down to the prompt because it's going to get a bad response back or like, you know, what things are you going to filter out at the last minute? because, like, it is a bad response that has gotten all the way through. You know, and sometimes, like, you have multiple round trips through this cycle before you, like, bubble the thing all the way back up to the user to get them the response that they need.

And so, like, you know, we have a point of view about what all, what this stack looks like, you know, which Microsoft tools exist that will help people build these things. And, like, what special things you have to go do in the context of an enterprise to, like, answer the actual direct question, where, you know, safety and data privacy and, like, understanding, you know, where the flows of data are and, like, which plugins can be enabled and, like, which can't, like, all of those things, like, I think are getting built out right now. And, like, the other thing, too, I'll say is, like, we'll build some of this stuff. And, like, the community is going to build a tremendous amount of it because, like, there's

never been a platform or ecosystem where one company builds all of the useful things. Like, that's just nonsense. Like, it's just never happened. And to me, it's the sort of super exciting thing to just see all of the energy that's happening right now. Like, I just, like, immediately before this call, I was doing a review with Microsoft research. And it's just amazing to watch MSR, which is so many researchers are there, have pivoted what they're doing research on to like these AI adjacent or AI, like, on point things.

And it feels a little bit like what MSR was like when I was an intern there in 2001, where, you know, you had all of these super bright people who, like, had the tiniest little glimpse of what the future must look like that no one else had because it was the point where the PC was racing to ubiquity. and like they were just all orienting their research around like what that little glimpse was that like maybe they had the earliest peak at and it just like feels magical that's massive realignment of the research community right now sort of in real time it's it's very exciting to watch i mean and it's awe inspiring i mean it's just crazy it's hard to keep up like super hard like we went from i mean this has been the biggest surprise for for me is like i just didn't realize that gp t and ChatGPT, we're going to catalyze as much of this as they have. Like, we'd sort of kind of been expecting a bunch of this stuff. You know, ChatGPT was a 10-month-old model with a little bit of RLHF on top of it. And, you know, like by, you know, admission, like, you know, not a beautiful user interface.

It was just sort of a way to get something out there because, you know, you needed some practice with a handful of things before the big GPT4 launch was. was coming and like no one really knew that it was going to blow up this way and it's only five months old that was only five months ago which is shocking i think everybody forgets how little time is past yeah just shocking but but it is the open source community and like the the you know big tech community i think it's best is like you know everybody is sort of realigning to like what i think is you unlike some of the other you know fadish things that have happened over the past a handful of years. Like, I don't think this is a fad. Like, this is, this is real. Yeah. I launched my new fund about six months ago with this AI focus, and a few weeks later, chat GPD comes out. And I'd say

even the people who are very prepared, like, hopefully somewhat prepared to go, like, try to keep up or be part of that massive shift, like, feel constantly upended. But it is, it is very, it's the most fun time to be in technology in decades. Yeah. Look, it's also, I will say, a disconnect. inserting time to be in technology because so many things are changing at once. It's changing at a pace that, you know, you probably, like even me, like I'm, I think I might be in one of the better positions to like feel like I'm kind of in control of what's going on and like I'm not in control at all, like of the pace. And so it must really be disconcerting to folks, you know, trying to keep up with everything that's going on. And in some cases,

is like it's forcing people to change their worldview about things, like worldviews that they've held for a really long time. I think it's honestly harder for some machine learning people than it is for like a brand new entrepreneur who's just looking for an interesting thing to go do because it is a very different way for a machine learning team to do its work. And it's like been hard, you know, even for some of the people at Microsoft who have had plenty of time to think about the transition to, like, get adjusted to, like, this new way of doing things. I want to ask you one more question that is sort of advice for people making the adjustment in a certain sense. And then, you know, talk about your book, talk about the macro and such. Microsoft has a unbelievably wide portfolio of products. And now you're on the other side of all the infrastructure questions, figuring out the, you know, organization of adoption of all these capabilities into that portfolio, right?

I talk to friends who run large companies, started large companies all the time that are also figuring out how to do this. How do you organize that effort? What advice do you have for them? I think you have to be, you have to remember that some things have changed and some things haven't changed at all. And so, like, one of the confusing things that I think there is for folks that many people get wrong is like models aren't products and infrastructure isn't. a product. And so, you know, you need to very quickly understand what it is this new type of infrastructure and this new platform is capable of. But that does not mean that you get to not do the hard work of understanding, like, what a good product is that uses it. Like, one of the things

I tell a lot of people is probably the place where the most interesting products are are where you've made the phase change from impossible to hard. So, like, something that, like, literally you couldn't do at all before this technology exists has become hard now. Because, like, the things that have gone from impossible to easy are probably not interesting. And, like, my frivolous example of this is when smartphones came on the market 15, 16, 17 years ago now. Like, it's, yeah, 2007, I guess, was iPhone launch, right? So 16 years ago, almost.

And then a year later, you had the App Store. So, like, the first apps were, like, things that had gone from impossible to easy. And, like, they just, you know, we barely remember them. Like, there were all these fart apps. There was, like, you know, like this app I had on my phone at one point that was called the Woo button. You pressed it, and it, like, did a Woo, like, Rick Flair. like those are those are not businesses like they're just you know sort of like these explorations that people are doing like the things that have made the smartphone

or the hard things that like went from impossible to hard they also are kind of the non-obvious things like they weren't even the things that the builders of the platform imagined like you know we we don't even think the the original applications on these platforms, like the things that launched, when the platform first launched, like, those are not the interesting things anymore. Like, your smartphone is way more than just an SMS app and a web browser and a mail client. Like, the thing that makes it interesting is TikTok and Instagram and WhatsApp and DoorDash and, like,

they were all of these hard things that people had to go built now that they were possible. And so, like, I think that's thing, number one, to, you know, hold in your head either as an entrepreneur or as a business that's trying to adopt this stuff. It's not like how I go sprinkle some LLM fairy dust on my existing products and do some stupid incremental thing. And I shouldn't even call it stupid. Like maybe the incremental things are fine. But like the really interesting things are non-obvious and very not incremental. And so that is the hard thing for us is you have an entire group of people who are smart and like they can see all of the things that are possible.

And so the challenge is to steer them towards, like, the hard, meaningful, you know, sort of interesting, non-obvious things that are possible, like, not the, you know, like things that are incremental that, you know, just going to burn up a bunch of GPU cycles and prevent you from, you know, in a bunch of product IQ that will prevent you from doing the things that really matter. If we sort of zoom out to, like, non-technical audiences, you wrote a book in 2020, reprogramming the American Dream. Can you describe who you want to read the book and what you hope they'll take away from it? I, when I wrote the book, it was not for people like us. So the premise of the book is that I grew up in rural central Virginia. My, you know, dad was a construction worker. His dad was a construction worker. his dad was a construction worker.

My maternal grandfather ran an appliance repair business and had been a farmer earlier in his life. So the thing that was true for everyone who was in my life, like, you know, neighbors, members of the community is like, you know, they're just smart, entrepreneurial, ingenious people using the best tools that they could lay their hands on to go do things that matter to them that, like, created opportunities. for them and, you know, sort of solve problems for, you know, their, their communities. And I believe that, like, particularly this platform vision of AI where it's sort of getting cheaper and it's getting more accessible all the time, you know, like things, you know, like this stuff that we were chatting about a few minutes ago about what I did at Google. Like, I, you know, came in with a graduate degree. I was mathematically sophisticated. And yet to do the thing that I, the first project that I did, which was a, you know, like a machine learning classifier thing in 2003, 2003, 2004.

Like, that was, you know, stacks of, like, super technical, you know, research papers and, you know, the elements of statistical machine learning, you know, like, you read it cover to cover, and then you go write code for six months. Like, high school student could do the whole damn project in four hours on a weekend now. Like, it's just, you know, like, and, and, and, and, and, and, and, you know, and, and, and, and, and, what's happening, like that aperture of who can use the tools is just getting bigger and bigger and bigger over time. And so, like, the book was trying to get people to be inspired by this notion that, like, don't be daunted and intimidated or scared by AI. Like, go embrace it and, like, try to plug it into the things that you're doing.

And, like, maybe, you know, we've got a shot at having more equitable distribution of, you know, who, who's benefiting from the platform as it emerges. If you were going to add an update chapter for the last few years where so much has happened, what would you focus on? Well, it's really interesting how much of it I think is still true. And like I had this anxiety the whole time that I was writing the book that I was going to, by the time I had the manuscript and it hit the presses, that all of it was going to be out of date.

Like the real problem I had is like by the time it hit the presses, We had a global pandemic, and it literally – it hit the presses the week that everything shut down, so, like, you literally couldn't buy it. Like Amazon wasn't delivering anything other than the essential packages, and every bookstore in the country was closed. So, I mean, it's a little bit surprising, you know, to me, like how many of the ideas that, you know, we have a platform. Platform's getting more powerful. It's getting more accessible. like actually the unit economics of it are getting better. You know, like what you can do for, you know,

per token of inference is like getting higher. You know, so like I know everybody's like in this frenzy around GPUs and like, which is this very expensive thing. But like all of this optimization work is happening where you're able to squeeze more out of the compute that you have and the compute's getting cheaper. So, yeah, I mean, the update that I would add is that, and it may be an update that I do, like, it probably won't be this book, but like I'm

sort of contemplating, like, writing something right now. I do think that the public dialogue around AI right now is missing so many of the opportunities that we have to go deploy the technology for good. Yeah, like all of the articles that you, you know, you read in the newspapers are, you know, around the responsible AI stuff, which is important. and like the regulatory stuff, which is important. But, you know, we should have a few articles in there as well about Sal Khan's TED Talk, which is just amazing, like unbelievably good. And just for, you know, folks who may not have seen it, which they should go see,

is like, you know, his problem is perfect for AI. So it's this two-sigma's problem, this idea that students who have access to high-quality individualized instruction performs substantially better than those who haven't, like, controlled for everything else. Just for our listeners' sake, the two-sigma problem was the study by a guy named Benjamin Bloom, which showed that your average tutored student performed above 98% of students in a control class, which is one teacher to 30 students, like a normal American classroom, with reduced variance, which is amazing.

Yeah, and if you believe that that's true, then you can also believe that. that every student, every learner in the world deserves to have access to that individualized high-quality instruction at no cost, which seems like a reasonable thing. And then when you think about how you go realize that in the world, like, the only way that you can realistically do it is with something like AI. And so there's so many problems that have that characteristic where we can all agree that Like, it is a universal good to do this. And then if you think about how to do it, like, you must conclude that AI is, like, part of the solution.

Like, that is the, you know, the reason I get up every morning and deal with people yelling at me about, like, give me my GPUs, you know, for the fifth year in a row. It is because of things exactly like that. And it doesn't mean that when you talk about that and you're hopeful and optimistic about those things or even hopeful and optimistic about all of the things that, you know, venture back companies are going to go do or, like, the way the businesses are going to reinvent themselves, that you are also saying, you know, given the middle finger to, you know, the responsible AI concerns or, you know, the things that people care about on the regulatory front. Like, you can care about both of those things at the same time. But, like, the thing that I can tell you is, like, there is no historical precedent where you get all of these beneficial things by starting from pessimism first. like pessimism doesn't, doesn't get you to optimistic outcomes. Yeah, it seems like to your point, a lot of the dialogue is really lacking from global education equity, global health equity, like all these things that AI as a platform should be able to produce because it's, it's cheaper, it's personalized, it can do things at the level of a human in many cases in terms of being a great teacher or a great, you know, physicians assistant, etc. And so it really feels like that message is lost. And, you know, I think a lot of people don't mention enough how we're, almost hopefully going to enter this golden age if we let this technology actually bloom and be useful. I guess the question that I always have on my mind relative to all this stuff is

given the capabilities that AI continues to accumulate, how do you think about 20 years from now in terms of the best roles for people? And in particular, I think about it in the context of my kids. I'm like, okay, normally two years ago, I would have told my kids, go study computer science. It's the language of the future. What do you think is the right advice to give people, you know, in terms of what to study and that will be the things that will be most durable relative to the changes that's coming. Yeah, I think, so 20 years is a tough time horizon, you know, and I think if any of us are honest with ourselves, like, if you rewind 20 years and you sort of imagine the predictions

you would have made then, like, would you have gotten here and like, nope, nobody would. But, like, I think, you know, there's just some sort of obvious things. Like, my daughter, for instance, like, has decided she wants to go be a surgeon. And, like, I think surgeon is, like, a pretty good job. Like, we do not have a, you know, sort of robotics exponentials right now. Like, we've got a cognitive exponential. And so, like, I think all of the, like, the world is just sort of full of these jobs where, you know, really, you know, affecting change on a physical system, like, doing something in the physical world, like, all of those things, like, we will need probably many, many more of them than we have right now, like, particularly in medicine. nurses, surgeons, physical therapists, people who work in nursing homes. We have a rapidly aging

population. And so, like, the burdens on the health care system are going to get much higher. And, you know, I do think that AI is going to have some pretty substantial productivity impacts. But, like, you know, maybe it's just enough productivity impact to, like, you know, make room for all of the other net new things that we will have to have there. You know, and so I think, You know, we got this weird thing in the United States where, like, we apportioned less dignity and respect to, like, jobs, like, the ones that my dad had, than we should. You know, and I lived in Germany for a little while, and Germany is a little bit different on this front. Like, you can, you know, you can go be a machinist in Germany, and, you know, like, that's a really great career and something that your parents are, you know, celebrate. So, like, I think they're, like, all of these careers, like, you know, electricians and machinists and, you know, solar installation technicians and, I mean, just so many things that we're going to need, like, especially because we're going to have to rebuild our entire power generation and distribution system, like, in our children's lifetimes.

So, like, all of those jobs, I think, are super important. And then I would argue even that all of the creative stuff that we do, there's going to be probably more need for that in the future than less, even though the tools that we're using to do the creative work, whether it's coding or making podcast or whatnot, are going to help us be better at it. And the reason that I say that is humans are just extraordinarily good at wanting to put humans at the center of their stories. So, like, we, right now, we could be, you know, making, you know, Netflix shows, you know, like not Queens Gambit, but Machines Gambit, like about, you know, a fleet of computers playing chess among themselves because they're all better than the very best human. nobody wants to watch that. Like, the technology is probably good enough right now where you could have superhuman Formula One closed track drivers in Formula One cars that could, you know, do things that humans can't do.

Nobody wants to watch that. Yeah, and like you even go back before computers, like, forklifts are stronger than people. You could, like, go have a strong man or a strong person competition that was about, like, which forklift could lift the most weight. Like, nobody cares about that. Like, we care about humans. Like, what are we? saying what do we care about like what are we trying to express to everyone else like and nothing

about that's going to change nothing i think that's why people watch the real housewives of dubai so yeah and and i don't again like i don't want to paint too rosy a picture uh every time you have a major technology platform or paradigm shift like there's disruption but like what we know from every one of these disruptions is you have, like, actually a surprising degree of need for human occupation, like, all of the, you know, the Industrial Revolution predictions about, you know, four-hour work weeks, and we're all going to live lives of leisure is bullcrap. Yeah, like, just hasn't happened. And I think some people may say it hasn't happened because, you know, the system, you know, the system, like, doesn't want it to.

happen, but, like, I think a lot of it is because, uh, like, we actually like doing things. Yeah. And there's a lot to do. I guess on that note, what, what are some of the areas you're most excited about going forward in terms of the coming year of AI or, you know, big research areas or big product areas or things that, you know, you're very optimistic about? I think sort of two things. Just, um, I think this will be the, maybe the great first year of foundation model deployments where you're just going to see like lots and lots of companies launch, lots of people trying a bunch of ideas, you're going to see all of the big tech companies will have substantial things that they're going to be building. I got predictions about what other

folks will do, but it will touch all of Microsoft's product portfolio. The way that you will interact with our software will be substantially different by the end of this calendar year that it was coming in. And I think that will be true for everyone. I think it changes some of the nature of the competition that you've got between big tech companies. And I think it creates new opportunities for small tech companies to come and drive wedges and to get footholds and do interesting things. One of the things that Sam Altman and I have talked about a lot is, I suspect that this year, like, the next trillion-dollar company gets founded. It won't be obvious which it is, but, like, we're overdue, like, long overdue.

And then I think, you know, what you're going to see technically this year is I do think that you will have things like the Red Pajama Project is like this, and they're going to be a bunch of others like it, will make really good. progress on building more capable open source models. And hopefully, hopefully the community will help build some of the safety solutions that you will need to accompany those things when you deploy them. But, like, technically, I think you're just going to see amazing progress there. And then, like, it's just, yeah, the frontier will keep expanding out. You know, we don't have, OpenAI doesn't have GPTV. wide distribution, but, like, it'll get to wide distribution at some point in the not

too distant future. And so, like, you'll have these, like, very powerful multimodal models the same way that having GPD4 admitted, like, all of this exciting energy around new things that you could do with it, like, having a, having a model that can, like, take visual inputs and, like, reason over them, like, will also admit a whole bunch of new things that are going to be very exciting so i don't know like that that's like i just think the the theme of this year is going to be like progress and activity like almost too much to track like i'm going to need a i'm going to need a co-pilot just to like pay attention to all of this stuff and like make sure i'm not

missing important things because i feel like i'm i'm at the and you all as investors and as people who are watching this closely must feel the same thing it's like um how do i make sure i don't miss like you know the next important thing how do i see it as soon as human possible. Actually, just to make completely sure, if you are starting the next trillion dollar company in our listener base this year, please call me in a lot. And Kevin, too. Wrapping up now, is there anything else you would want to touch on, Kevin? Look, I think the dialogue that we're having right now around regulation is actually really quite important. So as we're recording this, Sam Altman was testifying in front of the Senate

Judiciary Committee on like Tuesday of this week, I think more of those conversations are a good thing. I think as fast as things are moving, like you really will need the technology community to come together and to agree on some sensible things that we can do before the regulation even is in place. And I think that's all important and not a thing. Like, the thing that none of us should be doing at this point is sort of, like, looking at the prospect of regulation and saying, oh, my God, this is a, you know, this is like a pain. Like, I don't want to deal with this.

Like, the fact that there is a desire for it is, like, a very good signal that the things that we're working on actually matter because, like, nobody's trying to regulate frivolous things. And, like, the purpose of regulation is to make sure that you can build a solid, trusted foundation for things that, you know, become ubiquitous in society. Like, if you think of this, like, electricity, for instance, you want to strike the right balance between allowing the technology to develop and make progress and flourish, but, like, you also need to make sure that your electric power generators are built safely and you don't allow people to wander in and, like, stick their finger on the electrode and disintegrate themselves. And, like, you want to make sure that, you know, the distribution of electricity is coordinated and that, you know, when it comes into your house, like,

like it doesn't burn your house down, and when you, you know, like you plug your appliances into the wall that, you know, they function as expected. And so, like, I think that is, you know, a similar way. Like, there's not going to be one size fits all. Like, I think the most of the stuff that people ought to be thinking about is deployments. Like, making sure that, like, as you deploy the technology, getting, you know, the requirements and the expectations right there is the most important thing. And then, you know, these big engines that we're building that are the, like, the largest of the foundation models, like, you know, making sure that, you know, you have a set of safeguards around those.

But, like, also the way that we're building these things, they don't, they don't get distributed to the world, like, by themselves. Like, there's a whole layer of things on top of them to, like, render them safe. and then a whole set of things per application per deployment that we do to make the deployment safe. And so, like, you know, I think everybody, like all the startups, like everyone in the open source community, everybody ought to be thinking about these things. Like, how am I doing my part to make sure that we are creating as much space as possible for these optimistic uses? and like we are deterring as many of the harmful ones as possible.

Yeah, I've been impressed by the degree to which the community has self-acted from very early days in terms of AI safety and approaches to that. And so I know opening I has done stuff really early, Anthropic has, Google has, Microsoft has. You know, I feel like a lot of the main players have actually been, you know, remarkably thoughtful about this area and, you know, keen to make sure that it's done properly. Yeah, I mean, the thing that I will say is we fiercely compete. with a whole bunch of these folks. But, like, one of the things that I don't do is, like, look at any of those companies that you just named and, like, worry that they're going to do something that, like, is, like, I take myself out of my role as CTO Microsoft and, like, just think about Kevin's Citizen of the World. Like, Kevin's Citizen of the World is not worried about, like, what my competitors are going to do to, like, do something unsafe. Like, I'm just not.

Thanks so much for being with us, Kevin. We really appreciate it. Yeah, thanks for inviting me. This is awesome. Yeah, thanks so much for the time. That's great.