THE AI GAME CHANGER: HOW TECHNOLOGY IS TRANSFORMING BASKETBALL AND BEYOND
AHMAD RASHAD: So now let's kick things off with the first of our five panels. It's about something that is transforming sports and society at large, AI. And I'm not talking about Allen Iverson, as I said before.
As this exciting new technology revolutionized the world around us, we'll hear from some of the sharpest minds about what to expect in the future.
And here to lead this discussion is a brilliant journalist we know from "The New York Times" and CNBC. He's no stranger to moderating conversation. His annual DealBook Summit always draws some of the biggest names in government, finance, and culture.
Let's give it up for Andrew Ross Sorkin.
(Applause.)
ANDREW ROSS SORKIN: Good morning, everybody. It is a pleasure for me to be here with what a remarkable group of people to talk about a technology that is genuinely changing our lives mid-game as it's happening right now, but I think is really going to be changing our lives going forward.
Let me introduce to you who I am here with. This guy needs no introduction. Steve Ballmer, of course, the chairman and owner of the LA Clippers.
(Applause.)
ANDREW ROSS SORKIN: And I should mention, of course, he was the CEO of Microsoft from 2000 to 2014, but maybe more importantly, he joined that company in 1980. He was Employee No. 30.
STEVE BALLMER: Exactly.
ANDREW ROSS SORKIN: So we've got a lot to talk about.
We've got Fidji Simo, who is the CEO of applications for OpenAI. For those of you who play with ChatGPT, she's responsible for a lot of what you're seeing these days. She ran Instacart and before that was at Facebook. And I think the future of AI in large part may be coming from a lot of what she is up to.
Matt Garman, all the way on the end, is the CEO of Amazon Web Services. Amazon just announced in the past week that it plans to spend -- are you ready for it? -- $200 billion, with a B, in the next year alone, just on infrastructure to power the future of what is about to happen to us.
So I keep saying what's about to happen to us. Steve, I want to start there. And there's a question about sort of where we are in this whole world.
We're going to get into the sports of it all, because Adam has already referenced a whole bunch of things that I think are going to change, but when you look at just where we are in all of this, are we at the tip-off? Are we at halftime? Are we in the first quarter? I happen to believe it's possible that we're still just in the shoot-around.
STEVE BALLMER: I'm going to take it a bit further than that. We were in shoot-around before ChatGPT got launched. Right now, we are certainly before the first timeout in the first quarter, even for a team that's definitely behind, maybe two minutes into the game, in my opinion.
This is the most -- I've lived through a lot of amazing things -- the PC, the phone, the internet, et cetera -- this thing is 1,000 times more important. They were all free intelligence in a small form with a computer. The world's open to you. The world's information is open to you.
Now, real intelligence, I mean, it's stunning to me.
ANDREW ROSS SORKIN: What is the -- right now, in this moment, the most exciting part of it to you, and also maybe the scariest?
STEVE BALLMER: As we think about things, you know, what now is focused on is how AI helps people do things. And when the metaphor flips, which I think happens in the next year or two, the question will be: How do people help the AI?
People will back the AI as opposed to the AI backing the people. Whether you have a call center, or we're worried -- worried. We're thinking about applications to do financial education and counseling for low-income people. The only way to do that is have the AI be the front and the person be the back.
And I think that happens next year, year and a half. I'll look to people who are active in the field, but I think that'll be an amazing change.
ANDREW ROSS SORKIN: Okay. I want to get into what it's going to mean on the court and off in just a moment.
But, Fidji, help me with this. It feels like, and I just spent a whole bunch of time with some sort of super serious Ph.D.s, researchers in the past week or two that have been playing with things like Claude Bot and some of these other sort of new technologies, that literally in the past two weeks, I would argue, have fundamentally even changed where we even thought AI could be.
What does that look like to you?
FIDJI SIMO: So it looks very similar to what Steve is talking about, which is fundamentally we are used to AI in the form of ChatGPT, which is like you chat with it, it's a dialogue back and forth. It's really good.
But what has transformed in the last few months is really the ability to build with AI. And we're seeing that with Codex just kicking off. That's our product that allows you to build really without having to write a line of code. You're just prompting the AI to do something. And so we're really moving from a world where you were chatting with the AI to now doing stuff with the AI.
And I think the most fundamental thing is that it's going to remind all of us that, at our core, we are builders, we are creators, we are born to create. But a lot of people are limited by their technical abilities. Like to create things, like applications, things like that, you needed to code, and that was a very big limitation.
Now I think we're going to absolutely blow past that where everybody is going to be able to build, and we're going to see just an explosion of software. And I really agree with Steve. I think the world looks like everyone managing a team of agents.
ANDREW ROSS SORKIN: What does that mean? When people say they're -- look, we've got a whole room here. We've got athletes in the room, we've got CEOs, we've got people who own teams, we've got marketers. When people say that their agents are working on behalf of them, what does that mean in practice?
FIDJI SIMO: That means in practice, if you're a manager today, you are probably going to start your day telling your team, Hey, these are the five projects I need done, please go and do that.
In the future, you're going to tell agents that. You're going to go into a Codex app, and you're going to be saying, Hey, I need this full analysis done. Please go do it. I need to build this app. Please go do it.
And you're going to check throughout the day that they're doing their job, and you're going to guide them and train them, the same way you train a human employee. You're going to say good job, not so good job. And that's going to refine over time.
To me, like the best anecdote I found in the last week is we have one of our execs actually in the room, Kevin Weil, who came to a meeting, and he said, You know, like if I don't start a task through my agent before coming to a meeting, I feel like I've wasted an hour because I have all of this compute available, and now I can start, like, anything. And if I don't, I'm limited just by my own imagination.
And I think we're all going to be thinking that way: What are the agents that we can run in the background to do stuff for us so that we can create even more.
STEVE BALLMER: Can I give you an example of my favorite agent?
ANDREW ROSS SORKIN: Please.
STEVE BALLMER: My favorite agent is the agent that represents our player's agent. So if you want to have the negotiation with the player in advance, you need an agent for player X. And you can literally have the dialogue, the discussion, and try to practice what that player negotiation looks like by having an agent for the player's agent.
ANDREW ROSS SORKIN: Pretending to be the agent.
STEVE BALLMER: Exactly.
ANDREW ROSS SORKIN: When do you think that you're going to go from actually negotiating against the individual agent versus the digital agent? When the athlete says, You know what? I don't need an actual agent that's taking 10 percent. I'm going to take my digital agent, and they can negotiate with Steve Ballmer.
STEVE BALLMER: I think two things are true. Number one, we'll need more accuracy. Like, pretty close to perfect. And then number two, I still think decision-making is one of the few jobs that's absolutely secure in this world.
ANDREW ROSS SORKIN: You do.
STEVE BALLMER: Actually making final decisions I don't think are going to be trusted for a while, a final decision: Will we invest? Are we going to pay Charlie?
I think there'll be human intervention both on the -- let's call it the player side and the team side as an example.
ANDREW ROSS SORKIN: Okay, Matt's making faces. So I think he's got something a little crazy to say about agents. I know you guys use them.
MATT GARMAN: Look, it's true today, but it's -- you know, I think it's -- I think today it's actually very hard to say what won't be true because the technology is changing so fast.
And so, as an example, I would trust an agent at Amazon to decide if we refund $100 to a customer or not. So then now it's just shades of gray as to, like, what you're deciding it to do. And so today, I agree, I wouldn't sign a $50 million contract and just let the computer decide.
I do think, though -- and who knows if we'll ever get to that spot. If we will or if we don't want to, by the way. And I do think that one of the things that we'll -- that the world is coming around to and is going to kind of understand is what do we want these things to do?
And there are some of these, we may not want them, right? It may be that the player actually wants to talk to a person and that they actually care about the owner and they actually want to have that personal connection.
And so that actually may eventually be why Steve's right that you don't have an agent do it because the people actually -- there still are people behind these things.
And so some of these pieces, I actually think some of that person-to-person interaction are some of the things that will be durable for a while because people like them.
ANDREW ROSS SORKIN: Okay, let me try a different one out. Players using a digital coach. So right now, there are people, maybe even in this room, who have used ChatGPT almost as a psychiatrist. They've asked them questions about how they can be better, what they're doing right, what they're doing wrong, how they can react in a particular moment.
I imagine that if you get a bunch of good tape of a game, in the future, players, actual players, will feed it in and say, "Diagnose what's going on here, help me figure out -- by the way, we're playing so-and-so next week, really tell me about their defense," in a way that maybe the individual can't even see.
MATT GARMAN: Yeah. We're already doing some of these things. So we work with many of the leagues to do some of these, and with some of our partners, where we actually look at individual players, and we look at all of the data from an entire game.
And we watch on the NFL. Did you run this route exactly right? People look at it, they actually understand. Did you move right? Did you -- did you -- were you doing something that could lead to injury?
And players kind of ingest a bunch of this information in basketball. We're looking where we look at every second, we look at every data point of all the limbs of all the players to understand where they can optimize their game and where they can move more efficiently, where they can shoot more efficiently, where they're not going to the right place.
And if you think about from a coach's perspective, if you think about during a practice, it's hard to watch every player every time and tell them, Hey, you didn't do the exact right thing on this play.
But computers can do that, and actually it's a super valuable coaching tool because you can highlight the one spot where somebody didn't quite set the screen at the exact right place or where they could have moved more quickly to rotate on a defender. And you can highlight those and get -- and teams and players can actually try to get better using that.
FIDJI SIMO: And I think the thing that's interesting is overlaying that, then, with the fan data. We are seeing, for example, the San Antonio Spurs do that, NASCAR analyze tons and tons of fan data to understand, correlate that with what happened in the game to understand what are the types of moments that create enthusiasm for fans versus frustration. And then they use that to refine the rules of the game as a result, which I find fascinating.
ANDREW ROSS SORKIN: Okay, and there -- go ahead.
STEVE BALLMER: We've got just one thing that will help us with this. In our arena, we can tell whether you're in your seat, you're not in your seat. We can tell how loud you've been, down to the seat level. We can tell when you were buying something when you weren't in your seat.
And so the ability to take the kind of intelligence that Fidji just talked about, I mean, it's here, the stuff Matt talked about. Today, we already can give, first, our analytics guys the tools that say what's the best way to guard this version of this pick and roll? Great.
But now you want to give it to the coaches, you want to push it out to the players so they can ask their questions: Okay, Kawhi and Brook have to defend pick and roll against Houston. What's that really going to look like?
They can ask the questions, they can bring up the video, as opposed to having have the video guy cut it for them, so to speak.
ANDREW ROSS SORKIN: Okay, different question. It may be that players on the court are going to be one of the most protected professions, if you will, in the age of AI.
But I want to read you something. This is Microsoft's AI CEO in the FT this week says that: Most, if not all, white collar tasks -- and there's a lot of folks who work in the white collar world in our audience today -- will be automated by AI within 18 months.
Anybody want to react to that?
FIDJI SIMO: So the way I kind of think about it is like it's true that a core of the job -- and what I mean by that is like the tasks that people perform -- are going to be able to be automated.
But I think the ability to decide what to create, what to build, et cetera, you still need, like, a human spark. You still need impetus behind AI to figure out what to do.
And the thing I'm seeing is I'm seeing the best -- like I'm seeing a lot of companies say, okay, we're going to automate a lot of jobs. That's going to be great cost reduction. But the best companies are actually saying, no, actually, I'm going to get a 3X productivity gain. What are all of the things that I didn't think were possible that are now possible that I can create?
So I think we're also going to see massive growth. So that's going to be a disruption for sure. And that's why we're doing a lot of programs to help people upskill for these AI skills.
But I think we're actually going to see a transition where more jobs are also going to be created because people are going to be able to do so much more. So the ROI on one person is going to be so much greater because every person will be massively amplified by AI.
STEVE BALLMER: We already have that in basketball, by the way, in a small way.
ANDREW ROSS SORKIN: What do you mean?
STEVE BALLMER: We have a live game. People love the live game. You need live players. But there's also a video game that creates digital content. I'll bet if you were to add up -- and I haven't done that -- all the time that people take, you know, playing NBA 2K, what you have is a digital world, in which people are spending a lot of time, and a physical world.
So the question is, it's sort of Matt's point earlier, people will be able to have it both ways. To Fidji's point, wow, now we have new scenarios, new things to do.
So I'm going to say the video game, at least in live entertainment, is a precursor to all that you'll be able to do with AI digitally and with the most amazing athletes in the world.
ANDREW ROSS SORKIN: Matt, help me with this. If I'm walking into my meeting the way Fidji describes, and before the meeting starts I'm assigning nine of my agents to go do work that, at least historically, I would have assigned nine people on my staff to go do, does that mean that the nine staffers don't have jobs, or now those nine staffers are also using this to multiply themselves? How does that work in reality?
MATT GARMAN: Yeah, I think it's actually a fascinating and difficult question that all of us are going to have to answer over the next 18 months. I think --
ANDREW ROSS SORKIN: But it's that quickly. We're going to have to figure this out in 18 months?
MATT GARMAN: I think we probably have to figure it out faster than that. Like, here's the thing that I think we all --
ANDREW ROSS SORKIN: Wow.
MATT GARMAN: We should probably figure it out faster than that.
ANDREW ROSS SORKIN: Let's figure it out now. It sounds like we should just do it here.
MATT GARMAN: Look, there's no chance that 18 months from now every single white collar job will have been automated. Like, the world just doesn't move that fast. Will it be possible to be automated? Maybe. But it won't be.
And I think the thing that I think everybody has to realize is that we really have to internalize that every single job will change. The thing that you did that made you great at your job three years ago is very unlikely to be the thing that is great two years from now. And you pick the timelines.
But it doesn't mean that all jobs go away either. I think that in the software world, as an example, we have always been limited by the number of great people who could write software. There's always been more software that we've wanted to write than we could do.
Now, that equation may flip, and it may be that you're limited by good ideas of what to go build or how to build it. And it doesn't mean that spewing out more stuff is going to be better, by the way. It may have to be how does it hang together, how does it work together?
So there are going to be hard pieces of how this works together, but the jobs will change, for sure. And you're going to think about how you send jobs to agents and how you collaborate with those agents and put them together and put them in front of customers and drive value, and some jobs will go away.
ANDREW ROSS SORKIN: When you say put them in front of customers, most people here are used to putting a salesperson in front of a customer, like physically. What do you mean by that?
MATT GARMAN: I actually think sales is a super interesting example. So we have a very large sales force in AWS. I think we do a good job of being super efficient with that sales force.
And if I was to roll back two years ago, I think, best case, they spent 20 percent of their time in front of customers. The rest of the time, they're filling out their Salesforce records, they're building account plans, they're building presentations that they're going to present to the customer. They're doing a bunch of stuff like that.
Even if you can take small changes and you can say 60 percent of their time is now in front of customers, I don't think that part goes away. Because, again, the customers are going to want to talk to you. They like talking to chatbots, but they like talking to humans too.
And you can make that interaction more efficient. You can do a lot more things there. You can, realtime, kind of build demos for them, where you used to have to go back and write, build something, and come back. Like, there's a lot of better interactions you can have there.
But they're still going to want that human interaction to ask questions, and I just think that society is a long way from everybody not wanting to talk to people all day.
FIDJI SIMO: But I think the skills you hire for are going to change dramatically. I think the people who are thriving in this environment are the generalists who are like builders and tinkerers. It's people who really master AI skills. It's people with high agency that come up with those ideas.
So a very different world from, like, when we were hiring for deep specialists. In the past to build products, we had the product managers, the engineers, the designers. And what I'm seeing right now is people building stuff at OpenAI are all of the above. They are in AI tools all day long. They're coming up with the idea. They're coding, like using the tools, they're designing. They're doing all of that.
So my biggest piece of advice would be, like, hire for people with agency who are living in those tools and are going to be able to predict that future.
STEVE BALLMER: My biggest worry is that the education system will screw this all up; that the education system won't say use AI wherever you can, it will say no, we have to continue to teach you how to write.
Nobody needs to learn how to write in the classic sense, at least that I did. Now, I tried to pick something at the edge of provocative. But doing -- letting kids live early on the same way they have to live when they get into the real world. That, I worry a lot about that.
ANDREW ROSS SORKIN: Can I just ask a question about that.
STEVE BALLMER: Andrew is like whoa, whoa, whoa.
ANDREW ROSS SORKIN: I'm a professional writer that --
STEVE BALLMER: Yes, sir.
ANDREW ROSS SORKIN: -- I may be out of a gig.
STEVE BALLMER: I could have picked Matt, my stuff.
ANDREW ROSS SORKIN: Well, no, no, but this is interesting. So, you know, we have calculators, but we still teach kids math as a bit of a logic train, how to actually think. Similarly, I would argue that writing -- and, look, there was a time you actually had to know how to spell. You don't really have to know how to spell anymore.
But at some point, does the logic train break down if you don't understand the underlying thought process behind how we're getting to these outcomes?
MATT GARMAN: I actually think teaching writing is a super important skill, and I hope that schools continue to teach that.
(Applause.)
MATT GARMAN: Steve can go with PowerPoint, that's fine. But I do think -- I think writing actually is that underlying -- it is that underlying logic piece that actually I do worry that eventually you have to learn how to use all of the tools, absolutely.
But I think your calculator example is a very good one. I don't ever do arithmetic in my job, but I definitely need to know how to do arithmetic. I think it helps with your logic and learning and other things like that. But, yeah, I have Excel do arithmetic for me.
ANDREW ROSS SORKIN: What do you think? You think you don't have to know?
STEVE BALLMER: I'm being only slightly provocative. Whether it's today or tomorrow --
ANDREW ROSS SORKIN: I think you're right about what's going to happen.
STEVE BALLMER: It's got to be held -- you don't want to hold kids back. You don't want to hold kids back because the jobs they're going to find, you're not going to spend a huge amount of your time, mostly, writing. I mean, that's part of what we heard. You're not going to spend most of your time doing -- even building a financial model. You're going to do that thing with the tools.
And right now, if you look at, oh, no, we can't ever let a kid submit a paper with which they've used AI to assist them, I look at it the opposite way and say if we're going to equip people to do, if we're going to enable what Fidji talked about in terms of new creation, new things to do, we can't tell kids we're going to treat -- you know, train you like troglodytes and then drop you into the world.
That's what I have to say about that. And I do love your books.
ANDREW ROSS SORKIN: Thank you.
FIDJI SIMO: I have a 10-year-old daughter.
ANDREW ROSS SORKIN: Still no AI, but maybe the next book.
FIDJI SIMO: I have a 10-year-old daughter, and in one weekend, using ChatGPT, she was able to build an entire website, an entire set of marketing assets for a business that she created. That gave me so much more hope for a future than getting an A in math. So I'm with you.
ANDREW ROSS SORKIN: Okay. So all of this raises a very interesting question about the NBA and about players and about brands. And the reason I mention this is, you know, your daughter is building a website for herself in a weekend. And right now there are certain players that are spending a lot of time on social media.
They might have teams of people that are trying to help them build their brands. There's many of them have become entrepreneurs, but they have a whole sort of army around them to try to build that.
And the question is will they still have all those people doing that in the future? Will they just be the CEO of it all and have agents do it for them? And if everybody can do it, how does it change how the public perceives all this and how we think about the sort of power and influence of a particular player in this league? Does it become much more about the team? Does it become much more about the players themselves even more?
STEVE BALLMER: I think brands will matter even more. Brands will matter because people will need to trust, whether they need to trust the technology that they get from Amazon, whether they need to trust the validity of "That is Kawhi Leonard who I saw play last night." The sort of provenance and brand I think matters even more.
Now, will there be some automation of the workaround? Sure. This Tiffany Norwood synthetic actress, there will be synthetic basketball players who play in, you know, 2K leagues. It won't all be, you know, you're going to have, I don't know, Joe Schlomovovitz, and Joe is absolutely the best player right now in 2K. And then somebody will come around with the creativity tools and say, no, you know, I've got Javon Schlomovitz, his cousin, or whatever, whatever it is. People will create those kinds of things.
ANDREW ROSS SORKIN: And will people want to watch the bots playing each other? I mean, that's a whole separate matter because people talk about sort of AI slop. If they know it's made by AI, people don't give it the sort of credibility and credence necessarily than they do if they think that you, Steve, actually did it yourself.
STEVE BALLMER: I agree. And nobody's going to quite know the difference. When people go watch video games on Twitch -- I think you guys still own Twitch.
MATT GARMAN: We do.
STEVE BALLMER: I think when people go watch video games on Twitch, you can already argue whether they really know in many instances who's playing.
MATT GARMAN: Yeah. I think this is true, and I agree that brands are going to matter. I think, like, we have -- there's proof points of this today. Literally anyone can post a video on YouTube today. And you have folks that were out here today that have hundreds of millions and billions of views. I don't have billions of that posting on YouTube.
But there's lots and lots of content, and you still see a concentration of this is better by some definition of better. And it may be AI generated that's better, it may be person generated that's better, but I think that there will be a brand of something that is better.
I think one of the things that will continue to differentiate, whether it's entertainment content, whether it's business content, whether it's applications, is the data that is underlying these things. Because everybody will be able to go build software, but you have to have data that feeds that software.
And I think that is one of the differentiating capabilities for a long time, is do you have interesting, unique data that allows people to have -- that allows you to differentiate the thing you can build from the same thing that a 10-year-old can go build.
FIDJI SIMO: I also think that the creative tools are going to help with the fan engagement piece, because players cannot spend their entire time taking selfies and videos with their fans. But with a technology like Sora, you can put yourself in the field with your favorite players, and that's a magical experience for fans.
So I think that will also change how immersed into the sport fans can be.
STEVE BALLMER: Do you know how they have those cams in the arenas today that will focus in on a fan? We need to work to put that fan in the game.
FIDJI SIMO: Yes.
STEVE BALLMER: So instead of just seeing them in their seat, you can also see them running up and down the court --
FIDJI SIMO: I love that.
STEVE BALLMER: -- through tools like Sora. I'm kind of keen on trying to get that going on our big board at Intuit.
ANDREW ROSS SORKIN: Let me ask you a question. You guys talked about brands and the power of brands and the power of trust. But at the same time, if your agent is going out to buy you a product or get you something, how much do you think people are going to value the actual brands themselves, or if they're trying to get a service or something, versus just having the agents either do it themselves for you, or effectively go out and just find the most inexpensive, quality version, how that actually all plays?
FIDJI SIMO: I think for commodity products, it's definitely going to be much more about price and how fast it can ship. And, like, Matt would know more about that than me.
But I think for aspirational products, brands will absolutely still matter. But what can happen is that the agent can get to know you so well that they are going to be able to find brands for you that you wouldn't have naturally found and are the absolute perfect fit for your exact taste. And that I think is pretty magical.
ANDREW ROSS SORKIN: Okay, so now I got a really complicated one. And I know I want to talk about advertising, which gets to all this, and obviously there was the war of the Super Bowl ads in the past weekend between OpenAI and what Claude and Anthropic were doing. We can talk about that in a second.
But to the extent that everybody here has an agent, will we even really ever see ads, or will our agents see the ads for us? I know, very Meta here. Another technology brand here.
Didn't get a laugh. It's okay.
STEVE BALLMER: You got me.
ANDREW ROSS SORKIN: Yeah, that was like, you know, off the rim. It was a bank shot.
MATT GARMAN: Thought it was clever.
ANDREW ROSS SORKIN: No, but this is a serious question, in terms of you guys are going to be introducing ads, for example, on ChatGPT. Talk about what that actually even feels like, because I think there's some questions in the marketplace about what that is. But if in the future my agent is just going to go find stuff for me, I may never see those ads.
FIDJI SIMO: I think you will still have some interaction through the screen, even for what Steve was saying, which is you need to make decision. Like, you need to tell your agent what to do, decide if what they're doing is the right thing. So that's an opportunity to showcase to you a potential thing.
But I do think that we are going to live in a world where brands are going to need to care even more about the data and the reputation because agents are going to be able to make the best choice based on all the data online about a brand's reputation, the quality of their products, the reviews.
And so they're not going to be able to influence people just by putting an ad in front of them. They're going to need to actually have great products so that they get picked by the agent.
So I actually think it's going to raise all boats.
MATT GARMAN: Yeah. I also think that you have to think about what an ad is too. I mean, if you rewind 40 years ago and you ask people, like, is the single biggest advertiser -- advertising platform in the world gives you links that are blue links with little underlines under them, people would be like, That's not an ad. How is that possibly the biggest ad platform in the world?
But it is, right? And so, like, the definition of what an ad is can morph over time.
And, look, I think this is for sure a problem that -- this is one of the problems that the world will need to solve because people aren't just going to let that go away. They want to be able to drive traffic, they want to be able to influence what agents see, they're going to want to influence how people direct their agents, all of those type of things.
And so I think that you may have a tweak of net change in different ad models that someone hasn't invented yet, but I still think that there will still be a opportunity to pay money to influence where people go to purchase things.
ANDREW ROSS SORKIN: How much harder is it going to be to create a true breakthrough star or new brand in the future? Given -- I mean, we talk about sort of the fragmented world, and obviously you go back and look at Michael Jordan during a different heyday of the media landscape. You could look at LeBron today.
I think I would argue it's getting harder already. But I wonder in an AI universe, where you're talking about sort of an even more splintered world, unless you think sports is somehow different, it changes things.
STEVE BALLMER: I think live entertainment -- and I continue to believe there'll be live entertainment as well as virtual entertainment. I think live entertainment is different, but it's also different by sport.
Football is unusual already today. People don't necessarily recognize football players when they see them on the street. Basketball players, if you're one of the top, you get recognized. There's just no question. Soccer is a little more like basketball. And you can go sport by sport.
But the truth is, I believe that there'll still be fascination with human beings. Same, I'm sure, is true in music and many other fields.
Some of them are tougher, but at least for live entertainment, my belief is there's relative brand security, and you'll have even better tools for the best, the top guys, whether it's Cooper or whomever, to come out and really tell their story.
So I think that part, there will be famous people. As Matt said, there will be head-end content, which gets disproportionate amount of the attention. I believe that will continue to exist.
ANDREW ROSS SORKIN: What do you guys think? And the reason I ask is I imagine every ad that we see, to the extent we see ads, may be directed -- very specific and even different. I mean, that's sort of the beauty of what AI can do, is that every ad can be different for each individual person. And as a result, though, you may not -- if everyone's not seeing the same thing, it may be a different world.
FIDJI SIMO: Well, you can imagine both worlds, right? I think there will be both worlds, where AI enables much more personalization. So you're going to be able to have, like, extremely niche content, and maybe it's a favorite player just for you.
But I also think that because AI is going to lead to an even more connected world through agents, that, like, breakthrough can happen even faster.
I mean, if you look, like ChatGPT reached 100 million users way faster than all of the prior waves of technology. And part of it is because everything is a lot more connected.
So you could imagine, if there is truly a breakthrough moment, all of the agents being like, oh, everyone needs to know about that, let's go tell our humans that they should care, and that happening way faster in a connected world.
MATT GARMAN: I also think that there is some aspect of community in that too. If you think about sports, I mean, it's more fun when you're at a sporting event, when you're high-fiving and your team wins the Super Bowl last weekend, as an example. It's fun, right?
And it's not as fun as if, like, I have my own individual, whatever, NBA 2K team that won on my video game. Like, that was fun. I liked playing that game. But it's not quite the same thing as that community getting together. And I think those things are unlikely to go away.
STEVE BALLMER: When you go out with your basketball shoes on, you're still going to have somebody's brand on your kicks. You're still going to want to impress your friends. Man, I'm old-timer. Man, I've got a pair of Air Jordans. I just can't -- and maybe I'm too old-fashioned, I can't believe that goes away.
ANDREW ROSS SORKIN: Let me ask you guys a slightly different question, and it was actually something that someone in the back had asked me, which was -- we were talking about privacy. A lot of people are going to be putting a lot of information into these systems about themselves.
I just tested a thing where actually it got to read some of my e-mail. And then I asked it, What do you think of me? By the way, you guys might have done this on ChatGPT already. People say, What do you think of me? Be my biggest critic. All this kind of stuff. It's totally different when they actually have your emails because they can go back.
And I'll give you a crazy example. The ChatGPT said -- or the system said, Well, it seems like you sometimes prioritize -- my wife's not here -- you prioritize your work over your family, and there was a moment -- and I said, Why do you think that?
And it's, Well, back in December, your son was having surgery -- he's fine, by the way, now -- and you seemed to continue to be writing emails from the hospital.
And I said, Well, how do you know that?
And it said, Well, you sent some people -- you were talking about the hospital and all this, you kept doing work things.
Now, by the way, there's not much I could have done during the surgery.
But the idea that it could -- that it knew that and that it could directly sort of point to that, to me suggests a lot of things that we all as a culture and society need to think about.
So I'm curious how you guys think about that, and also how much private stuff you today are willing to tell any of these models.
FIDJI SIMO: So I tell ChatGPT everything about my life. But we take this incredibly seriously. I mean, what you're saying is already happening. This is the technology that is, I think, the most personal, the most intimate.
People talk about their mental health. They talk about relationship with their spouse. So we take that incredibly seriously. That's why we have a lot of privacy protection.
And we are also adding additional privacy protection for really sensitive data. We just launched ChatGPT Health, where you can now connect your health records to get a really accurate picture of your health, which is incredibly helpful, but incredibly sensitive data. So we put that in an even more locked-up location so that, like, the rest of ChatGPT cannot see that.
So we're really thinking through that because, fundamentally, that will change the relationship that people have with technology to have something so personal and private.
ANDREW ROSS SORKIN: Matt, what are you doing?
MATT GARMAN: I was going to say, we've actually thought about this from -- this goes back to at some point it is all about this data, again. And whether it's your personal data or your company data, I do think that having a good framework for keeping that data safe and secure, while getting value out of it at the same time, you can't just lock it in a box and never use it, is incredibly important.
And so our view for our business is more focused on the business world and how -- but it's the same thing. The business -- when everybody has infinite access to these tools, the IP and the data that you have is going to be what differentiates one company from another company.
And making sure that you have that data in an incredibly secure place -- and that's why from the beginning, in Amazon AWS, we've built this secure platform to ensure that your data stays there, to ensure that your data stays inside of your private network, and that you have full control over that. And I think increasingly it's super important for everybody to think about your personal data, your work data, and how you control that.
And, look, I think mistakes on that -- everybody is super aware of this, and it's one of the things that could slow this technology down through regulation or other things like that if there's mistakes made. And so it's incredibly important, and it's one of those things that if your data and IP gets out, you can't get it back.
ANDREW ROSS SORKIN: What are you putting in there, Steve?
STEVE BALLMER: Me, I'll put almost anything in. Now, there's two reasons in which that's the case. Professionally, I just do. And because of the capabilities of Microsoft Copilot, I can say things like: Hey, make me a timeline of how we made this decision. What did I have to say about the decision?
I mean, in the professional context, the kind of stuff that you talked about is doable today.
Now, when you get to things that are attorney-client privilege or otherwise needs to be in the provenance of a company, there's got to be the right tools. And Microsoft, I'm sure a bunch of other people, are working on those.
My personal life, I have a choice. If I want to, I'm me. And if I don't want to be me, I can be johnqrandom@random.com, and that can be my identity.
So I have tools where I can talk about myself, I can provide my various -- not all personal information. I probably can't provide my W-2 and still preserve confidentiality.
But then I got -- I got -- now, W-2 is not a crazy one because if you have people who don't have a financial advisor, who don't know how to budget their finances -- yes, I would encourage people, but these guys have to have enough -- and if you're talking about lower income people, which we focus on in our philanthropy, you know, you got to get there.
And if I'm not confident putting my stuff up there, why should somebody else be confident? So I've been pretty forward about this with ChatGPT.
MATT GARMAN: And to that thing, people shouldn't be afraid of it, right? People have put their W-2s into Intuit for a really long time, and you just have to have the right controls over those things.
FIDJI SIMO: We also have a feature --
STEVE BALLMER: That would be Intuit, the TurboTax product, not the Intuit Dome. We don't keep anybody's tax records.
(Laughter.)
ANDREW ROSS SORKIN: On that score --
STEVE BALLMER: Come on, you fed it to me, man, I had to hit it.
ANDREW ROSS SORKIN: That's baseball, dude.
Thank you very, very much for a fabulous, fabulous, fabulous conversation. Thank you, everybody.
(Applause.)
AHMAD RASHAD: All right. We're going to take a short break. I mean short. If you need a snack, head out to our hospitality area, which you passed through on your way in. Restrooms are in the lobby just outside the theater doors. Just make sure to be back in a couple of minutes.
FastScripts Transcript by ASAP Sports