Microsoft Research Podcast - 081 - Game on with Dr. Chris Bishop and Phil Spencer

Episode Date: June 19, 2019

Dr. Chris Bishop is a Microsoft Technical Fellow and director of MSR Cambridge, where he oversees an impressive portfolio of research including machine learning, AI, healthcare and gaming. Phil Spencer is the Executive Vice President of Gaming at Microsoft where he oversees everything from the design of the next Xbox console to the creation and release of blockbuster properties like Halo, Gears of War and Forza Motorsport. These two powerhouse executives are pushing the boundaries of creativity, technical innovation and fun across the spectrum of gaming genres and devices for nearly 2 billion gamers around the world. On today’s podcast, Chris and Phil discuss their respective roles in Microsoft’s gaming ecosystem, revealing a sort of “enrichment pipeline” that flows all the way from researcher to developer to gamer. They also give us an inside look at the close collaboration between the world-class research organization of MSR and the world-class gaming franchise of Xbox, highlighting Microsoft’s unique ability to deliver the tools, talent and resources that fuel innovation and help shape the future of gaming.http://www.microsoft.com/research

Welcome to the third episode in our recurring series of two-guest podcasts. This episode is all about games. Join us as we follow ideas from the far-out what-if world of MSR's gaming research to the far-reaching here's-what world of Microsoft's Xbox gaming franchise and see from the perspectives of both a research executive and a gaming executive how ideas collide, worlds come together, and everybody has fun. You're listening to the Microsoft Research Podcast,

a show that brings you closer to the cutting edge of technology research and the scientists behind it. I'm your host, Gretchen Huizenga. Dr. Chris Bishop is a Microsoft Technical Fellow and Director of MSR Cambridge, where he oversees an impressive portfolio of research, including machine learning, AI, healthcare, and gaming. Phil Spencer is the Executive Vice President of Gaming at Microsoft, where he oversees everything from the design of the next Xbox console to the creation and

release of blockbuster properties like Halo, Gears of War, and Forza Motorsport. These two powerhouse executives are pushing the boundaries of creativity, technical innovation, and fun across the spectrum of gaming genres and devices for nearly 2 billion gamers around the world. On today's podcast, Chris and Phil discuss their respective roles in Microsoft's gaming ecosystem,

revealing a sort of enrichment pipeline that flows all the way from researcher to developer to gamer. They also give us an inside look at the close collaboration between the world-class research organization of MSR and the world-class gaming franchise of Xbox, highlighting Microsoft's unique ability to deliver the tools, talent, and resources that fuel innovation and help shape the future of gaming. That and much more on this episode of the Microsoft Research Podcast. So we have a dynamic duo in the booth today for a special two-guest episode of the Microsoft Research Podcast. I'm sitting here with Chris Bishop, who is a Microsoft Technical Fellow

and Director of Microsoft Research in Cambridge, UK. And I've got Phil Spencer, who's the Executive Vice President of Gaming at Microsoft, and he leads the company's business across all gaming devices and services. That's right. Yeah. And I am the duo part. Chris is dynamic. I am duo. Well, welcome Chris and Phil, Phil and Chris to the podcast. Yeah, it's awesome. Thanks. I'd like to start on sort of a philosophical note to give us a framework for our topic today from the cultural

historian Johann Huizinga, whom I share a last name with, he famously said, play is older than culture. And Uber nanny Mary Poppins from your hometown, Chris, famously said, you find the fun and snap the jobs again. So we seem hardwired to play. From each of your perspectives, what's so important about games or play that a company like Microsoft would invest so much, both in research and product teams, in this area? Chris, maybe you could start us out. Sure. From a research point of view, games offer an amazing environment in which to develop new machine learning algorithms and techniques. And of course, we hope in due course, those new

algorithms will feed back not only into gaming, but into many other domains. But beyond the sort of the very technical machine learning techniques themselves, there's also the fact that gaming is an environment in which we can explore the relationship between AI and people and see how that can work in partnership. So it's a very rich environment in which to drive new research ideas. Phil, what about you? I think adding on that, what I would say, as you said, we look at play as an innate human need. It's something that we all do.

Microsoft's mission statement of empower every person not only is what we build, but actually finding audience for the things that we build. So to me, the nice thing about gaming is it is so pervasive. People play on any device. You know, give me five minutes and a piece of paper and I'll get bored and start playing tic-tac-toe with myself or something. And I think as we look at the application to empower everybody, finding human uses of technology that really can reach anybody, regardless of demographic, regardless of where you live, regardless of socioeconomic situation, I just think really opens up the opportunity for us as a company.

Yeah. I have to ask at the outset, what's your personal relationship to gaming and why are you interested in this area personally? I think, yeah, my all-time greatest gaming experience was actually playing a fantastic game called Portal with my sons and really discovering that amazing social connection that it brings. It's a very collaborative game, puzzle solving together, and just great father-son experience. Of course, that said, they're a little bit more experienced than me with the Xbox controllers. They're kind of about three orders of magnitude faster than me, but it was still good fun. They have an unfair advantage. It's true. I mean, for me, I teasingly say

that I think it's the only job that Satya would hire me for at the company. I don't think I'm kind of equipped to do anything else. But I think Chris hit right on it. Some of my best experiences with my family have been around play. I have two daughters. Some of my best friendships have been around this opportunity. I started playing with my father.

He was an engineer, brought a PC home. And when I look out there today, there are so many forces in our lives that try to create division between people, the othering of people. You're a this, I'm a that. And this is an art form in gaming that has a natural capability to bring people together in a cooperative social way to either win the round, to compete in a nice way with each other. And I just love that. I love that I can work in a space that has this ability to bring people together. And it's the thing that drives me kind of personally in this space. Chris, you were on the podcast earlier this year, and we didn't talk much about gaming. Turns out you're overseeing some pretty cool research in gaming applications,

which we'll talk about in a minute. But first, tell us why gaming is a sweet spot for the Cambridge Lab and what your vision is with gaming technology research. That's a great question. I mean, the Cambridge Lab has been around for more than 20 years. And throughout that time, we've been very fortunate in having an incredibly close collaboration with Xbox. And that's led to a number of pieces of technology actually getting out there into the real world. And you know one of the factors I think is just that we have a lot of enthusiastic gamers in the lab and you know in research part of what we're supposed to do is to come up with new ideas. We're not directed, we're supposed to be kind of

spontaneous and think a little bit out of the box and so people naturally combine their passions, they're excited about technology like machine learning and many others, and they're excited about games, and they want to bring those two together. So I think that's been one of the factors for sure. We're really, I think, uniquely placed in having this great research strength, but also being really the only company in that space with a gaming division. And so it's just kind of an unfair advantage we have, and it'd be crazy not to take full advantage of it. Phil, you have a classic version of the mailroom to boardroom story here at Microsoft. You started as an intern and now you're the EVP of gaming. Tell us how you've seen the company's position in gaming evolve over the years and give us your vision as to where you see it going in the future. Yeah, I started at Microsoft in 1988 as a programming intern from the University of Washington. Go Dawgs. Go Dawgs. And it was interesting, like many things that Microsoft started that were new to us back in the day,

some of it was driven out of a paranoia of what competitors might be doing. I mean, Xbox was really born out of this fear that somebody else was going to create a family room PC, that if we should lean in and we should go do that as a company. The thing I love about where we are in gaming today, we sit on the senior leadership team, Satya's senior leadership team, or one of the key pillars of the company, is let's aspire to. How do we show empathy? How do we empower? How do we use Microsoft as a platform for us and the things that we want to do?

And I love that drive as a company. And I just think we've made it through the, oh no, somebody else might do it, to the thinking about what we might be able to do and us aspiring on our own. I'd say the partnership with Chris's team and Microsoft Research has been so critical to that. And it's been nice to see how the company's rallying around the interest that Microsoft All Up has in the gaming category. What's fascinating is I've got a classic research person here and a classic product person.

And I don't know if it's historic or traditional separation of streams, as it were, don't cross the stream. But research and product have tended to stay in their lanes. That said, I think there's actually quite a lot of collaboration going on that people maybe don't know about, and it seems to be gaining traction now. So do you see a sort of enhanced momentum between the academic and the industry teams? And if so, why do you think that is? Chris, maybe you could start. Yeah, I think there's always been great collaboration, actually, if it hasn't always been so visible.

But I think there is a sense of sort of more collaboration than ever. And I think partly it's just sort of the pace of innovation. And I think there's also more excitement in research, particularly these days, around seeing the fruits of that research really impact the real world. And, you know, one of the great privileges of being at Microsoft is you can do great research, and then you can get that research out to hundreds of millions of people. And that's very tough to do that in a university environment, for example. And so I think, you know, more than ever, there's a real enthusiasm about very close collaboration. Yeah. How about you, Phil? From the product perspective, what's important about

collaborating with these guys? Well, we probably don't talk about it publicly enough, but Chris's team is our secret ingredient. When we look at the competition in our space in gaming, the decade plus that we have working together with Microsoft Research and the Xbox team in gaming around making it more fun, making it easier to find people that you're going to have fun with, safety and security online, something we're focusing a lot on right now. I just think it's so critical to get people who have a sole focus like the MSR teams do on these categories present a problem in areas where we're not actually sure what the solution might be or even what it might entail. And to kind of check in every so often and see the progress that they can make, I'm really both in awe of the team and the capability that Chris has and the impact that

they have on the business that we're trying to drive. Well, let's talk about some of the specific examples of what happens when research and product come together. I've had some pretty cool folks on this podcast, and they've brought some pretty cool folks on this podcast, and they've brought some pretty cool research to the gaming world. Lennon Ravindranath Sivalingam has talked with me about Hype Zone, and Nakunj Raghavanshi has just been in the booth to talk about Triton. And I encourage our listeners to hit up those podcasts because it's a deeper dive on the research.

But Phil, for a second, from the product perspective, talk about how things like Hype Zone and Triton are impacting your world. Yeah, Hype Zone, I think, is a great example to center on. We have a streaming service called Mixer. There are hundreds of millions of people now who watch video games getting played. And I know to some that might seem strange. I kind of go back to the birth of like ESPN, where you were going to have a television channel that only had highlights of sports and like, nah, that'll never work. Things like Twitch, things like Mixer are examples of this for the next generation of people coming up where it is their sports center. It's the place they go, the personalities they follow.

So we have this service called Mixer, and it was, again, one of these what-if scenarios. Well, how do you know what stream to watch? How do you know where the most interesting things are happening? You know, Hype Zone, we kind of borrowed the name from Red streams across the games that have HypeZone. And they're actually looking for the most interesting moment at that time live. And you can sit on the HypeZone channel and literally we will swap to the most interesting instance of that game on our service. And it is about bringing the fun to somebody in an easy way. But also when you think about the inroads that we make and you start thinking about the applicability beyond just game streaming, what about news on the internet? Do I follow one news service or do I have something that knows what I like and is able to watch all the streams and bring those to me if it's recommendations on something like YouTube or something? I think there's a lot of applicability for this kind of tech, which in so many areas and the things that

we invest in gaming is great. I think we're always looking at how it applies to gaming, but then also how it applies to the planet. Chris, how is it that you sort of think ahead of the curve, as it were, before it hits product? What kind of research paths do you follow? Because this could have fruit and this doesn't. It's got to be difficult. It's a great question because research is sort of this combinatorically infinite space of possibility. And the question is, you know, where should you look? And I often say to researchers, you know, the biggest part of the job is figuring out what research to do.

And then the rest is kind of doing it. It's almost not the easy part, but, you know, doing the right thing at the right time, not going out to something that's so impossible or so far out, it won't ever be real, but also just steering clear of kind of the incremental stuff, trying to do the thing which kind of sounds impossible and a couple of years later you actually land it. That's sort of the sweet spot.

That stimulation of, you know, brainstorming about the impossible and getting ideas and thinking, well, hey, maybe that isn't so impossible. Now I've got the kind of an idea I could have a go at that or something like that. That's actually very stimulating for researchers in terms of thinking about what they should be working on next. One of the biggest challenges in online gaming, and in pickup basketball if we're honest, is finding players that are at your own level because everyone wants a good game right and apparently algorithms can help us find similar players talk about this

new skills ranking system called true skill Phil what is it and where is it being used and how's it working out so I'm gonna expand that a little bit because I think the fundamental issue that you talk about really in any activity is how do I find the group of people that I'm going to feel most comfortable with and have the most fun or even achieve the most if it's something that's more work related. True skill for us in the pure gaming space is like having the AI analyze how somebody plays and what their capability is. And then when you go in to play another game and you say, I want to have a good game, I want to play against people that I'm fairly evenly matched with.

We're able to use our understanding of your capability and put you in a match with similar skilled people. But at the same time, it's not just the skill of the player. It's what about what language these people use when they play? Are they aggressive players? Are they not aggressive players? There's hours of the day, even something as simple as that, that I want to play and I don't want to play. True skill starts with your capability and builds from that. But the layering on on top of this, of the social dynamics that happen online that aren't just about the skills someone has. But also when I'm there, I feel like this is an environment where

I can show up as my true self. I can be most authentic and have a good time at the same time. And when I think about that, that applies in so many places in our world today, right? It's the social interactions that we have on so many of these social networks and other things. I think the work that we're doing together in that space is awesome. And I think there's just so much opportunity there. Chris, could you add to that from the research perspective? Because he's talking about AI algorithms that are watching me play. Sure. I mean, it's a super interesting piece of technology, TrueSkill,

because a lot of the work we do in machine learning uses techniques like neural nets and deep learning, and they're very sort of data-hungry, and they're amazingly powerful techniques. But TrueSkill is a really interesting showcase of a different approach to machine learning called model-based machine learning or probabilistic modeling. And really the secret there is that you, instead of having this big sort of black box that you train up with lots of data, instead you actually describe the process by which the data comes into being.

You say, well, here's one person, they play against another person, and there's a game outcome, and that's the data we observe. And then you run some, you know, complicated code corresponding some beautiful math called Bayes theorem. And it takes all the information you have about players and their skills, and it absorbs the new information of the game outcome and then kind of updates this skill evaluation. And it's all done in a way that takes into account uncertainty. So probability is sort of at the heart of all of this. And it's just one of the first really large scale applications of this kind of machine learning. It went out first in Halo. It's now part of Gears of War 4. And we're now thinking about how we make this very generic so it can apply to a really broad range of games. Well, Chris, aside from gaming, other researchers that you're working with are bringing gaming technology to healthcare.

And two more guests on the podcast, Cecily Morrison spearheaded Project Torino, which makes a programming language for children with visual disabilities. And Haiyan Zheng was recently on talking about several projects, including Physio, which makes a video game out of laborious cystic fibrosis breathing exercises for kids. Talk about the vision behind the research that crosses these borders. I mean, this is such an exciting area. I think we're really just scratching the surface. I mean, projects like, I mean, this is such an exciting area. I think we're really just scratching the surface. I mean, projects like, I mean, you mentioned the, we call it Torino internally, but now we've done a royalty-free license to an organization that's going to produce this as Code Jumper. I mean, it's just a beautiful idea of allowing initially kids who are blind or partially sighted to learn to code by having effectively each line of code is represented as this very tactile plastic object

that you connect to other plastic objects using wires and then you press a button and the code gets executed. But actually the really nice thing about this is that it turns out to be a great way for kids who have full vision also to learn to code and it's very inclusive because you know the blind kids and the sighted kids can get together and work together collaboratively to design code and change code and so on. I'm just really excited that's out there in the world now. Yeah. And what about physio? I mean, when Haiyan was on the podcast, I was like, that's incredible.

How did that find its way into research? You know, sometimes it's really hard to kind of trace the sort of ancestry of these ideas. I mean, you know, research is a... If only you could. If only you could, yeah. I mean, research, you know, think of it a building, you're going to put coffee in one end and sort of ideas come out the other, and there's this mysterious sort of organic process that happens in between. I love that visualization. Yeah, right.

More coffee in one side, innovation comes out the other. Chris's tricks right there. It is actually the fuel for creativity. Absolutely. Yeah, when both coffee machines break, it's a major... Innovation stops. It is actually the fuel for creativity. When both coffee machines break, it's a major... Innovation stops. The lab director has to rush down and fix the coffee machines otherwise. That's the most critical person on the team. Oh, man.

The guy who can fix the coffee machine. Well, along those lines, Phil, millions of people saw and were deeply moved by Microsoft's powerful Super Bowl ad this year about making gaming accessible for everyone. The tagline, when everyone plays, we all win, is perfect because it's true, disabilities don't reduce our desire to play. So tell us the story behind the new Adaptive Controller for gamers with disabilities, because this is another area of your oversight, which is the hardware that people use to play. Yeah, I love the story of the Adapt controller, not just because of the product that we were able to build, but really the team that started this. It started as a hackathon project inside of

Microsoft, people coming together on their own time, not all from the gaming organization. Again, similar taking a, well, what if we tried to help in this space? There've been a lot of third parties that had done some form of accessible controllers. Prices were very high. There were no standards. So we took the approach, if we could create a piece of hardware, basically sell it at what it costs to build, make it accessible to any other platform. So it's out there for Windows. It's out there for Xbox. We've sent it to every other competing gaming platform saying, hey, here's a base to go use. But the team after the Hackathon project

did this in the evenings, got it to a point where they brought it to our game leadership team. We looked at the product. We weren't sure how we were gonna be able to fund it. How do you market that? What is it gonna mean when it comes out? When you start that journey with that community, it's not a one-step journey. So if we're in this, we're in it for the long run.

It's not a PR beat for us. It's a product beat. And then it was really nice in the holidays to see, one, when it first shipped in September, how the gaming community worked through some of the agencies that really helped us with testing and feedback to put it out. And then there was a Microsoft ad that they put out with Owen, who's the star of the ad. Great kid. And you just see him light up. But I will say, you talked about when everyone plays, we all win.

The moment that gets to me the most, and I hear this so many times, is when Owen's dad is talking about Owen playing video games. The truest representation of Owen is when he's playing online, he's just Owen. He's not in a chair. He can decide how and when he wants to share who he is because to everybody else, he's a good gamer. And I love that.

I think the line the dad uses something like, when he's online playing, he's just like everybody else. And I hear that so often in our space that the more we make it accessible, it just brings more people in. Everyone can play. And something Chris hit on, which I also think is really important for us on the product side to think about, almost every instance where we've approached making our products more accessible, whether it's to sight, hearing impaired people. We make the product better for everybody. Yes.

It's actually not an either-or decision. So it doesn't surprise me at all that programming techniques that we would put in place for somebody who's sight challenged becomes accessible to other people. We've seen that. We've seen that in our text-to-speech work that we've done on console. And it's such a truism we have to believe that when you make things easier for somebody, you literally make it easier for everybody. Yeah. And in fact, Cecily was talking quite extensively about the fact that you can have situational disabilities. You may have two arms, but you're holding a baby and doing something with your phone. You're suddenly a one-armed phone user. Yeah. And things like that. The curb cut for wheelchairs is also good for cyclists and it goes way broader. That's right. That's right. It's just expanding the aperture of how you're thinking. And you can't always do that because your own situation doesn't always give you the life experience. So, Chris, on that same line, what's the vision in terms of this accessibility

thread, kind of adding on what Phil said from your perspective? Right. So as I think, you know, Phil hit on something really important there, which is it's a great frontier to drive research. Take to the next level. How are we going to get technologies to help people in social situations? And of course, that's incredibly important if somebody is blind. So it's harder to get all the social cues, eye contact and that sort of thing. So can technology help them? We've been doing a lot of research about how we might push the frontiers there. And then of course it turns out that now you take a situation like meetings, you know, video conferences, where again, like you said,

we all have disabilities at different times of the day in different situations. And when somebody's on the other side of the planet and so on, there are barriers to normal communication. And then some of the research ideas apply equally in some of these situations that you might encounter in an office setting, for example. Well, some of the gaming advances we've talked about, and we've even alluded to it already here in this conversation, have come out of the annual company-wide summer hackathon. But there's a more compact version of this held for Microsoft Research and Microsoft Gaming, and it's gained quite a bit of traction. What do you see emerging from this hack event for researchers, designers, and game developers?

Chris, why don't you start? Yeah, I think of it as sort of, you know, TED conference meets hackathon. It's a kind of a hybrid thing. You know, we've run a couple so far. The first one was quite conference-like and really sort of allowed everybody to get to know what was going on in the different teams but the second one was really high energy and it was very focused on putting together teams from research from gaming and just working together on crazy little projects i mean one little example was

some developers from the gaming teams working with researchers using machine learning to take visuals from games and classify the genre of those games. Just a little exercise in machine learning and a great opportunity for people to get their first experience of doing hands-on machine learning in this very collaborative environment. But I think the real benefit that we've seen are the new relationships that form of people getting to know each other and then going on to form new collaborations. So, Phil, how about from your perspective, what is this annual hack event or now semi-annual?

Yeah, every six months. That's semi, right? I think we can't name the conference because we use a copyrighted name for the name of our little... Internally only. That's right, internal. Maybe I've seen it on a t-shirt somewhere, but we will not sell those.

I love the fact that the teams have so much energy around this This isn't something that I dictated or I think chris dictated that must happen It's self-driven by the teams getting together many people flying in which is time away from friends and family to sit down and really just have interesting conversations about what opportunities are in front of us and then actually start typing and try to make some progress on certain things in a short amount of time, which is the kind of hacking nature of it.

We started off with the idea we're going to do it once a year. I think after the first one, there was this we can't wait 12 more months to go do it again. So the team self-organized again to go put something in place in six months. I don't think we'll get to monthly. I think we're going to try to pin it at six months. But there's just a ton of energy around those moments in time. And as we know, the work, everybody's connected now. So that work happens even between the times where we get the teams together.

But it's literally we have to close the doors because it fills up really quickly. But it is taking our gaming teams, not just our devs on the gaming side, our game design people, our artists, bringing them in with our researchers and categories. We usually pick a theme and they go to a secret spot and they just kind of have some fun, laugh, and also think about what if opportunities. And I love how it just builds on itself with the momentum the teams have. Chris, another researcher at Cambridge, Katja Hoffman, who was also a guest, has been spearheading some AI research on a platform called Project Malmo, which is based on the wildly popular game Minecraft, which you, Phil, I think brought into the Microsoft fold.

That's right. And it uses reinforcement learning to train agents. So how's this platform enhancing the gaming experience now? And how might it advance general human AI collaboration in the future? PAUL LEWISOHN- Right. So Minecraft is, you know, thank you, Phil. That was an awesome decision there. MELANIE WARRICK- Awesome get.

PAUL LEWISOHN- Yeah. Minecraft is fantastic. One of the great things about Minecraft is such a flexible environment. You can create the simplest little world up to something which is almost like the real world, and everything in between. So from a search point of view, you can kind of titrate the kind of level of challenge that you're presenting your AI. If

it's too easy, you just make it harder. And so last year we used Minecraft under this Malmo umbrella to launch a competition to allow researchers around the world to experiment with AI in the domain of reinforcement learning. And it's been very, because of this sort of open nature and collaborative nature, it's just brought lots and lots of people together. In fact, even within the company, I mean, it's been a great collaboration amongst the different MSR labs. It's not just Cambridge and Katja who does a fantastic job of leading this, but, you know, our labs in Redmond and Montreal and just bringing researchers across Microsoft internationally together, along with the great folks in Minecraft. So I want to shift a little over to the fact that we live right now in a fairly console-centric

world, but I'm told the future of gaming is anywhere with anyone on any device. So what role does cloud computing play in the gaming world? What can it do for both gamers and game developers? Yeah, in today's world, there's a little over 7 billion people on the planet. About 4 billion of those are connected to the Internet in some way, whether it's a mobile phone, a PC, a console, any other device. Half the connected world plays video games today. And if you think by 2030, analysts are saying four out of the five billion connected consumers at that point will be playing games. Not everybody considers themselves a gamer, but you give them 20 minutes on a bus

with a phone, they're probably going to start playing something. So when we think about our ambition as Microsoft, what scale should we think about? We don't think it should be about what device you own. We put the player at the center of the decisions that we make and we say, how do we bring the games content that you want to you on any device? How does your community move with you from device to device? And this happens on any other, you're a Netflix customer on your PC, on your phone, on your television, anywhere you go, you can log in and use Spotify, Netflix, any service, Facebook, Twitter. All of these things meet you where you are. Gaming is kind of the last bastion of device-first consumer engagement.

And as Microsoft, we say we won't accept that. We want to make you the owner of your experience. And you should be able to go from device to device. The issue we have is not all the devices are technically capable of displaying some of the rich worlds and characters and stories that our creators create. So cloud allows us to do the rendering of that experience in our data centers and then bring the video stream down literally to any device that can decode a video and take some input and allow people to play anywhere. And I always think this isn't really about people who say, okay, I'm going to play while I'm away from my PC or away from my console. Like, yeah, those will be users. But I think about people in emerging markets where their only compute device will be the phone in their pocket. But the connection of the internet

means they know what a Halo is, right? They know what a Minecraft is. They know what our brands and what our stories and characters are. And we're pushing to bring those experiences to the devices that they already have in their pocket. So, Nikunj was just talking about the secret sauce of pre-computing and having Azure manage that because it's just so expensive in CPU time

to do these big calculations that you need for realistic games. So from your perspective, Chris, how does the cloud help out the work you're doing upstream of the device delivery? Yeah, I mean, the cloud is such an amazing innovation, sort of the elasticity of computation, the ability to effectively put supercomputer, super, super hypercomputer capabilities in the pockets, as Phil said, of somebody in Africa, effectively. And this is sort of inconceivable in the pre-cloud era. So that scale of computation is incredibly important in the back end of research because it's driving really the state of the art in machine learning and AI. It's also really important for these sort of rich simulated worlds. I mean, one of the great technologies that comes out of the gaming world is very

rich simulated worlds, which can be used for other kinds of things as well. Reinforcement learning for autonomous vehicles or whatever it may be, or applications in healthcare and so on. So the cloud really plays a very fundamental role in the AI revolution very generally, I think. So the mission of Microsoft is to, quote unquote, empower every person and organization on the planet to achieve more. And it's an interesting context putting it into the gaming world. How do I achieve more in gaming? I win more, I get a better social experience. There's all kinds of interesting threads on that that you guys have addressed that I actually never thought of, which is fantastic. How is gaming and the technical research behind it important to the company's mission?

And what is it about gaming and gaming research that empowers people? One really interesting thing about AI and gaming from a Microsoft perspective is that it's not about how do we get a computer to beat the human, right? How do we achieve human level performance in game X, game Y, game Z, as if a human being was somehow a list of capabilities and you're crossing them off one at a time. This is, you know, this is a sort of much richer view of AI as a collaboration between people and machines effectively. How can the AI enhance the human experience in this case, you know, create more fun, create more rich, immersive, interesting, varied gaming experiences. So in terms of empowering people, it's really part of that AI narrative of the complementarity between AI and people. And it's true today, and I think it'll be true for a very long time to come,

that the capabilities of the machine and the capabilities of people will be different and they'll be complementary. And they come together in this very nice synthesis in the world of gaming. Phil, talk to us about your side. I really think about the word empower and what does that mean to a person. And I don't think it's Microsoft's job to define what empower means. I think about our work in the gaming space. It can be messy at times, like the world can be messy. There's violence.

There's people who play too often. There's the monetization's people who play too often. There's the monetization, how that works. One of the easy things that we could do as a company is lean back and say, okay, it's too messy. We don't want to play in that space. But then when I look at our mission statement of empowering every person, empowering every person's also going to be a little messy every once in a while. And I love that we learn and get the feedback through the activity of being relevant in a space where people are gravitating. I think how many kids on the planet, their first compute experience is picking up mom or dad's iPad and playing Minecraft. And what responsibility do we have as the curators of Minecraft in that loop with

somebody, the social interactions they're going to have when somebody comes into their world? What should that feel like? How does a five-year-old feel empowered in that situation? And what do we teach both ourselves and the people in that environment at those influential times that will permeate how people work, how they interact with each other in different social environments, whether it's physical or digital. I love the opportunity to empower. And I just think the gaming space gives us this opportunity to empower people to achieve more. And maybe the achieve is kind of satisfaction and fun, but also to tackle some of the issues that are out there

that are Microsoft scale opportunities. And the interaction right now between the company's values and principles in this space and what we want to stand for in the gaming space, I don't think have ever been stronger. And I think it's just an amazing opportunity for us. So I ask all the guests that come into this booth some version of the question, what keeps you up at night? So what should we be thinking about in the worlds that gaming at Microsoft, there are two sides to anything that you do. What content should we approve? What activities on the platform are safe for others? How long should somebody play? Who gets to control that? how they spend. As I said earlier, I want to be part of that conversation, not outside of it. If we decided that we wouldn't want to do gaming anymore, gaming is not going away. So the thing that motivates me is I'd rather be at the table and part of the conversation than

just sitting back and pointing at it. Because as we started at the beginning of all this, play is an innate human need. People are going to play. And I think we have a unique opportunity to be part of that. And part of that means dealing with the hard issues and the sticky issues and the things that maybe keep us up at night sometimes. But I'd rather be relevant in that conversation and part of it and bring a Microsoft point of view than outside of it, just pointing at it as something that's too hard for us. I won't accept that. Chris, what do you think?

What keeps you up at night? I'm what you call a technology optimist, right? I really believe in the amazing power of technology to do all kinds of great things in the world. And particularly with technologies like AI and really machine learning, they're ubiquitous. They'll be used in many, many different places. But when you have a very powerful general technology, as we know, there are ways that it could be deliberately misused. But also there are just ways in which you could accidentally kind of get it wrong.

And we've even seen in the early phases of this revolution situations, not too much in gaming, but areas where people, very well intentioned people, have hit some little bumps in the road, perhaps around biases and just unintended consequences of learning from data. But I think the good news here is that we, particularly Microsoft as a company, but I think we as a community are very much on the front foot. We're recognizing that there are these obstacles, these challenges, and in many cases, they actually become research problems and they're very actively being researched.

So again, I think once we recognize that we can be on the front foot and really try to avoid most of those. Well, this has been a fascinating conversation and it's not quite over yet because I have one more question, but I want to thank you guys both for this eye-opening Venn diagram overlap descriptions of this world. So here's your chance to talk to would-be researchers and developers who are also gaming enthusiasts out there. I love this quote from Satya. He says, we don't want to be the cool company in the tech sector. We want to be the company that makes other people cool. I don't know if he actually said that, but if he didn't, he should have. What's on the horizon for the future of gaming and why is Microsoft

a good place for the cool kids to be. So I actually think of three things here that I'd just love to see come out of this. I mean, the first one is seeing this amazing collaboration really driving the frontiers of AI. We're seeing this already, new algorithms, new techniques for machine learning being driven by the challenges that arise in the world of gaming. So that's the first one.

And then the second one is to see the flow of ideas in the other direction, to see that we're really impacting the world of gaming, creating rich new experiences, and just do wonderful things for hundreds of millions of people, if not billions of ideas in the other direction, to see that we're really impacting the world of gaming, creating rich new experiences and just do wonderful things for hundreds of millions of people, if not billions of people around the world. That's an amazing opportunity. And then even the third one is even more ambitious to see this amazing collaboration lead to things that spill over and spin out into other fields. And the healthcare one is a particularly exciting one. So we're really quite ambitious.

Phil, how about you? Yeah, I think that maybe we can be the cool kids and make others cool at the same time. The opportunity to put our customer, whether the customer is a developer or the people using our products at the center is really what I hear from Sati. It's not about our gratification in the work that we do. It's what impact does that have on others? I think that is such a pivot for us. Microsoft for a long time has been centered on what we do and maybe how we do it.

Now we're getting more and more focused on why we do what we do. But I love that transformation that I've seen inside the company. And I think so much of the work that's going on right now just has application that we see today, but also application that we don't see tomorrow.

And that's why I think the ongoing work for all these developers on lean in and learn, growth mindset, I can kind of tick off our cultural principles. But I think that we have such a rich place to go learn from

and some incredible people to learn from, like Chris and Hyann, the team at MSR is just awesome. And I love that opportunity. It's such a great thing about working here. And it's just a great chance to say a huge thank you to Phil and your amazing team. And I, you know, it's been a fantastic collaboration. I really just love to see this carry on for many years to come. Great. Thanks, Chris. And as it turns out, cool is not a zero-sum game. We can be cool. You can be cool. We can all be cool.

That's right. And you guys, you guys are really cool. So thanks to both of you, Chris and Phil, for making time to share your insights on the podcast today. Great to have you both here in person. Thanks for the questions. To learn more about Dr. Chris Bishop and Phil Spencer, and how Microsoft takes playing games seriously, visit Microsoft.com slash research.