AI on the pitch at the World Cup

Note: Transcripts are automatically generated from episode audio, and are not fully corrected for spelling, grammar, and formatting.

ARTHUR HU: Think of it as the brain and the nerves of how FIFA run the World Cup, because they can’t have any downtime. Can you imagine if you are in the middle of a match and suddenly the broadcast feed cuts out or the replay system doesn’t work? We can’t have that. We have a lot of unsung heroes delivering that both to the fans in the stadium as well as online. Our partnership with FIFA is about helping bring the right technology into the game. Lenovo are not referees. We don’t make the rules.

RANA EL KALIOUBY: So it’s not like an AI referee that’s going to take over making these decisions?

HU: FIFA have been very clear they want this to be an augmentation. The star is not the system. The star is not the replay. The star is the competition on the pitch and what the players are doing.

KALIOUBY: Art Hu is the chief information officer of Lenovo. You might know Lenovo as the world’s largest PC manufacturer. This summer, they are the official technology partner of the 2026 Men’s World Cup, and Art and his team are responsible for deploying AI during live games inside the stadiums and to enhance the fan experience. To do that, Lenovo is working very closely with all 48 teams, all 16 cities hosting games across three countries, and the media companies bringing the World Cup to an estimated six billion viewers. Lenovo is basically defining a moment in history that will separate pre-AI sporting events from post-AI sporting events at an international scale, and I have so many questions.

I’m Rana el Kaliouby, and this is Pioneers of AI, a podcast taking you behind the scenes of the AI revolution.

[THEME MUSIC]

Hi, Art. Nice to be connected.

HU: Thanks for having me, Rana. It’s a pleasure.

KALIOUBY: Thank you for coming on the show. Before we start, are we going to call this soccer or football? I think we need to agree on terminology here.

HU: I struggle with this because I grew up in the US, so I used to call it soccer. But it’s weird. I call it football now because I’ve walked into the FIFA headquarters enough times that calling it soccer just feels weird, so I’m going with football.

KALIOUBY: OK, maybe we’ll go with football for this conversation.

HU: And if we slip, that’ll be OK.

KALIOUBY: Exactly. I grew up in Egypt, and I have the fondest memories of our family huddling around in the living room and watching the Egyptian team play. We weren’t very good.

HU: But it’s the country team. You have to root for them.

KALIOUBY: It’s the country team, totally. My most vivid memories are of my dad and my uncles yelling at the TV and all the women sitting and praying. There was a lot of praying. How about you? Did you grow up watching a lot of football?

HU: I grew up watching American football, but I think this has always been on the radar. In the last few years, as Lenovo has worked with FIFA, I’ve developed a much deeper appreciation for the game and all aspects of it, including the beauty of the competition on the pitch, but also a much deeper knowledge of what it takes to actually pull off one of these events.

Copy LinkWhat it takes to power the biggest World Cup ever

KALIOUBY: This is the largest World Cup ever. As we’re recording this conversation, just a week or so before it launches, how are you feeling?

HU: Excited and terrified at the same time. We’re T-minus six days from the opening match. We are locked and loaded and making the preparations so that we can make it as smooth as possible.

KALIOUBY: How will you be watching the games, and who are you backing?

HU: I mentioned I grew up in the US, so just as if you grew up in Egypt, you’re going to root for the home team, I’ll be doing the same. I will actually be doing most of the watching from the other side, because Lenovo, as the technology partner, is with FIFA a lot behind the scenes. My team is helping deliver everything behind the curtain, so we’re in the broadcast center, in the stadiums, and at headquarters. I’ll have a unique vantage point heading into this World Cup.

KALIOUBY: What is different about this World Cup specifically in terms of how AI is going to impact everything at the stadium and on the field, for fans who are watching live or at home? Can you bring it to life for us?

HU: We’ve got more than 17,000 Lenovo and Motorola devices. We’ve got more than 200 engineers behind the scenes. For example, we’ve got ThinkSystem servers at their international broadcast center. I was at the kickoff with President Infantino, the Dallas mayor, and the city council. They were feeding thousands of screens. We’re working on their IPTV latency so that it can be much more real time at all the sites that are relevant for them. We have also got a command center in Miami, which is the tournament operations center as well as the technical command center. Think of it as the brain and the nerves of how FIFA run the World Cup, because they can’t have any downtime. Can you imagine if you are in the middle of a match and suddenly the broadcast feed cuts out or the replay system doesn’t work? We can’t have that. All of the preparation and the discipline means we have a lot of unsung heroes who are going to be behind the scenes, end to end, providing what FIFA needs as they produce the match and deliver that both to the fans in the stadium and online.

Copy LinkWhy Lenovo saw FIFA as the ultimate stage for AI

KALIOUBY: How did Lenovo get involved in this?

HU: This actually goes back to some discussions several years ago. Lenovo is a very different company than we were 10 years ago. People often remember us as, “Oh, they sell wonderful laptops, ThinkPads,” and those are world-famous, but we have so much more now. We are a full pocket of cloud. We have Motorola Mobility, we have servers, we have storage, we have managed services, as well as professional services. We do all of the things that you need for an enterprise to run. We were thinking, “How do we get our story out there?” We really landed on FIFA, and there was a high resonance of, “Wow, they really are literally the world’s biggest stage, as close as you’ll get to an event where everyone’s eyes in the world are on them.”

As we started discussions with them, we found that there was a real shared ambition. When you walk into the Zurich headquarters, they have “Football Unites the World,” and they really mean it. Of course, it’s around the World Cup, but they have youth leagues, they’re always scouting, and they have member federations from countries all around the world that they’re trying to support and promote soccer through. They are highly ambitious about that mission, and a huge part of that is using technology to improve access, improve the quality of competition, and make that available to more and more people. The resonance with Lenovo is that we want to have smarter AI for all. We also have an ambition to bring AI to all parts of the world. As we got into that, we saw that it would be really interesting because they were very interested in using technology to advance the level of play. That matches very well with Lenovo’s desire to apply AI and the latest technologies at the cutting edge.

Once you find that common ground, the rest follows very easily. Then you can start to ideate, and you can talk about some of the possibilities. We found a lot of interesting shared ground that we could work on. That really formed the basis of the partnership. It’s not just a sponsorship. This is actually about deeply integrating and partnering with FIFA in all aspects of what they think of as the football family.

KALIOUBY: In prep for this interview, I reached out to Bibo, my brother-in-law. He lives in Cairo, and he is a super-diehard soccer — football — fan. He’s actually part of a group called The Football Freaks, and they watch all the games together and kind of ideate and pontificate. He was so excited that you and I were having this conversation, and he shared all the things they’re wondering about. I would love for you to step back and tell us all the different places where AI will show up this year at the World Cup. Then we can unpack each of them separately, but let’s start at a high level.

HU: Absolutely. I’ll try to give the bird’s-eye view, and then we can pick which ones to go into.

One of the things that we are doing is around 3D avatars. Right now, literally as we speak, in the run-up to the opening match, we are in the process of scanning each and every one of the 1,248 players who will be participating on the official roster, and that’s really interesting because it has multiple uses. On one hand, it’s going to be something that feeds the officiating and analytics systems so that we can support the referees with better data than they’ve ever had to make their calls. What I mean by that is, if you think about the video assistant referee, the VAR, and the semi-automated offsides, historically, when you see it on the screen, it has a generic avatar of a player relative to the offside line and how that’s playing out, but that has limited accuracy because it’s a generic avatar.

With the scans, we’re going to be able to provide data for the officiating and analytics systems that are true to the actual person as scanned, so things like what size uniform they wear, what their boot size is, and how that will show up. That increased level of accuracy will really help that part of the system. The other aspect is that it’s actually a better experience for fans, because they’ll be able to see it more clearly. There is a replay when the system comes on, and it’s used to show that additional accuracy and level of realism. So that’s one example.

Another thing that’s very important is something we’ve built with FIFA called the FIFA AI Pro. It’s built on Lenovo’s AI factory, but the point of it is to think about having elite match analytics and elite access to the best coaching can offer. Because FIFA are the football experts, we are bringing that football expertise and knowledge and training a model that’s going to be available, and is already available, for all 48 teams who are participating. This really democratizes access to elite coaching, and it raises the floor by providing every participating team, regardless of budget, resources, or how many coaches and assistants they have, access to more than 2,000 metrics for every match.

KALIOUBY: So cool.

HU: And it does really interesting stuff. It really is access to coaching, and that can be match strategies, tactics, 3D visualizations, and simulation. That one was really exciting to build because it’s not just functionality. It’s actually fundamentally different in terms of providing capability that we think democratizes access to elite analytics and AI-driven coaching. And that’s exciting ultimately because the more access teams have to this, the better the matches will hopefully be, and that’s really what’s exciting. That’s what we’re here for.

I’ll pick one more. As a technologist, I like doing things that are firsts. Some sports have an action view, but you can think of it as a relatively fixed camera view. But when you put a camera on a referee, they might be looking at the sky, they might be looking at fans, they might be looking at the ball, or maybe they’ve tripped over something and they’re on the ground with the players. The ability to stabilize that and provide a better sense of what it’s actually like on the ground is going to be so immersive, and we’re doing it in a way that is going to be available for broadcast in real time.

Copy LinkHow AI can improve officiating without replacing referees

KALIOUBY: So exciting. Okay, so let’s take each of these one by one to dig in a little deeper. The first one is around the AI refereeing and the officiating. So VAR, the video-assisted reviews, has been around since, I think, 2010, when it was first introduced?

HU: Yeah, it’s been more than a decade, yes. VAR by itself is not new.

KALIOUBY: Yeah. Is AI going to change this process this time around?

HU: I want to be very clear: our partnership with FIFA is about helping bring the right technology into the game. Lenovo are not referees. We don’t make the rules. The original video assistant referee didn’t replace anything, but it was an enhancement that provided more context and more ability for the referees to do their jobs as laid out by the rules. I think the latest enhancements, in terms of having the 3D avatars, follow in that lineage. We are not looking to, and by no means are we, changing the rules, nor are the referees doing anything different because of the technology itself. It’s really much more about AI as an augmentation of their intelligence, so they can have more fidelity, more granularity, and additional context so that they can make the most accurate call and the fairest call as described by the rules.

KALIOUBY: Yeah. Okay, so it’s not like an AI referee that’s going to take over making these decisions?

HU: FIFA have been very mindful of that. I can’t speak on their behalf, but they’ve always been very clear that they want this to be an augmentation to the capabilities they already have. The reason they do that is that we want the fan’s attention to be on the competition. The star is not the system. The star is not the replay. The star is the competition on the pitch and what the players are doing.

KALIOUBY: Yeah. This is actually something we talk about a lot on this podcast. A lot of my perspective on AI is that it should augment human abilities and not replace what we do or replace us as humans. I really do appreciate the approach where the AI is a tool that augments the decision-making of the experts who have been doing this forever and know what to look for. It’s just helping them do it in a better way.

HU: Yeah. I think the passion is really fun to watch because there is nothing that can replace that. Making sure we continue to have the human element at the heart of that, including the refereeing, while making sure they’re at the pinnacle of what they can do, is really our North Star.

KALIOUBY: Yeah. That’s another thing that’s come up a lot on the show, this idea of cognitive offloading. The more you offload to AI to make a decision on your behalf, the less capable you are of making that decision yourself. Is that something you guys have thought about?

HU: Yeah, definitely. I think both as a technologist and with FIFA, they’re very mindful of that. I would make a slight amendment to your statement, which is that if you are mindlessly doing cognitive offloading, then I think that is a very real risk. Your own skills may not be as sharp; they may deteriorate. If you just offload everything, whatever you offload may atrophy. So you really have to be thoughtful and say, “Well, it’s okay because I still know how to do that, and I’ll be able to spend my time on the next most important thing that either I couldn’t do before or I didn’t have as much time for.”

And so this goes back to how we are supporting the referees. This is not a replacement, as we said, but it provides them even more context at better speeds so that they’re able to make the call and, hopefully, do so even more accurately. That means they can do their jobs better, and then it goes back to keeping the focus on the gameplay. We have to be cautious that we are not blindly offloading as much as we can, because that’s not the thoughtful, measured approach that leads to good outcomes.

KALIOUBY: More with Art in just a moment after a short break.

[AD BREAK]

Copy LinkThe technology making offside calls more precise

I would love to unpack the technology behind the semi-automated offside technology. One of our producers is a soccer mom, and she’s very passionate. She basically said, “Offside” is the most common thing yelled at a soccer match. So now you’ve got these avatar scans, or these 3D scans, of each player, so you’re able to track exactly where each player is. What’s the technology that’s doing that?

HU: Yeah, so there are many cameras in the system. You can think of this as an add-on and an enhancement that the digital avatars are built on. FIFA work with Hawk-Eye to have the base system, and that’s not going to change. The core technology there, with higher-resolution cameras being able to track that, remains in place. We’re bringing in an additional data stream so they can overlay the higher-fidelity avatars versus a generic model of a soccer player playing in the World Cup. So that is the marriage of a core process and technology that already exists.

KALIOUBY: I’m just thinking it through. How would you know that it’s player X who just crossed the offside line? How do you marry the 3D avatar with the Hawk-Eye camera system?

HU: Yeah. A lot of engineering goes on in the background to get from the raw scan of the 3D models to being able to re-render them in the right graphics pipeline and input them into the existing systems so they can be accurately displayed. These are digital assets that also have a life cycle, so we have to make sure that they’re properly secured. We talked about the high-resolution raw scan, but it doesn’t do us any good if, when we re-render it, there’s a variance from what we scanned. There are a lot of quality checks to make sure that we have the scan, we’re able to process it, store it, secure it, and use it at the right time with the right accuracy, because the core algorithm is already there. You can look at what the plane is, according to the rules, for what offside would be, and this just makes sure that when we render the avatar, it is absolutely faithful to the high-resolution scans that we’ve made.

KALIOUBY: Yeah. By the way, who owns all of this data? That’s a treasure trove.

HU: Yeah, it is. FIFA are the ones responsible for capturing it. They’re the ones who have access to the data, and that’s also a good segue to what makes FIFA AI Pro so powerful. As we talk about democratizing the game and providing elite-level match strategies and match analytics to all the teams, that is what makes it unique. Another way of asking the question, especially when we were building this, is, “Why can’t I just go to your favorite LLM and ask, ‘Hey, who’s going to win? Analyze this match for me’?” And you can, but then you get problems where, A, it’s not accurate, and B, it’s not grounded in expert football knowledge and language. All of that data that FIFA has unique access to, from the highest level of competition in the world, includes thousands of metrics per match. Many of those metrics are around where the ball is, what force it is under, what the spin rate is, what its position on the field is, and how high it is going, just to name a few.

They have thousands of metrics, many of which are being captured hundreds of times a second, and so they have this explosion of data from the highest level of competition that we can use to train the model supporting FIFA AI Pro. And what’s nice about it is that, because everything is moving toward more real time for FIFA AI Pro, these are things that are going to be updated as each match proceeds. There’s no way any general consumer app can do that. So every match that gets played then helps the model get even more insightful for the teams that are using it.

Copy LinkHow FIFA AI Pro gives every team elite coaching insights

KALIOUBY: That is so cool. So let’s talk about how coaches and teams are actually using FIFA AI Pro. Are they using it during match play? Are they using it as they prep for a match? Are they using it to debrief after a match? What are some of the use cases?

HU: Yes, all of the above. Because this is the first time this is being used at a World Cup, everything you said and more are use cases, because we’ve made this available to all of the teams. There are no restrictions, first of all, on when they can use it, and it supports strategic analysis. So if you want to prepare for a match, you can ask, given these players and this coach, what is the likely strategy based on how they’ve played before? You can use it for player analysis: “In which positions has this player been most effective?” You can use it strategically or tactically. How compact do I want my defense to be if I’m trying to defend against three back versus four back? So you can basically ask it any question, and it’s very finely tuned.

As you can imagine, the way match analysts think about it versus fans is very different. They have very specific language around how they describe this. They can use it before the match, during the match, and after the match. Preparing for any of those scenarios, and all of them, are ones we’ve suggested to the teams as available. The beauty of it is that the rubber is meeting the road. They now have it for testing as each of the teams is arriving and preparing. Because we see football strategy change so quickly, teams want to be very adaptive, so they’ll be able to use this if something is emerging and really working well. We’ll be able to tease that out and help teams analyze it, as well as improve the functionality. To your point, we’re not restricting the teams because this is the first time we are trying this. We anticipate getting a lot of really interesting feedback about how to make the functionality better or support more scenarios that maybe we hadn’t thought about.

KALIOUBY: Yeah, I was going to ask about that. There’s a lot at stake here, so how do you ensure that the model doesn’t hallucinate in terms of its recommended strategies?

HU: Well, this goes back to our earlier point. This is why we don’t have robots playing the matches with a robot coach and a large language model behind a screen with a smiley face coaching. The human element has to be at the forefront. I think, at this point, the technology is well understood enough. Even if you look at all the consumer and enterprise tools, every single thing says, “May or may not be true, please check.” You kind of laugh if you think about it, and the same thing applies here.

Now, we are very confident. No LLM-based system will be 100% accurate, and I always like to zoom out and ask, “Well, when was the last time a human was 100% accurate?” But culturally, we accept it. Sometimes I screw up spellings, Rana. You’re not perfect. We accept it. It’s OK if you’re wrong 1% of the time. But somehow we have in mind that, because it’s a machine, it should never be wrong. We have very high accuracy, just from the testing we’ve done. We think, and are highly confident, that it will be right most of the time, in terms of delivering accurate information based on what it’s been trained on. At the same time, it still has the asterisk: “Please make sure you check.” I think this is part of how it will be used.

KALIOUBY: You know what would be fascinating? At the end of the day, these models are generative, right?

HU: Correct.

KALIOUBY: I wonder if they’ll come up with really creative strategies that a human coach would be willing to take a risk on and try, and then we would see what the results are.

HU: It absolutely could, and I think that’s one of the things to watch for. We’ll see, so stay tuned. It’s part of the mystery and the excitement for this World Cup. It’s another variable that could make match play more exciting.

Copy LinkHow AI can spark creative strategy

KALIOUBY: We’re going to take a quick halftime break. Don’t go anywhere. I want to bring it back to what you said about FIFA AI Pro democratizing access to football and football analytics, and that’s really amazing.

HU: I think what’s really nice about our partnership is it’s not just that. There’s so much focus on the players, the coaches, and the teams, and rightfully so. But pulling this off is also a Herculean feat. One of the other areas, back to the theme of behind-the-scenes heroes, is this intelligent command center. Remember, to watch the players on the field, they first have to get there. I was just in the operations control center in Miami, and the intelligent command center we’ve helped build with FIFA is actually helping them. Think of it as a giant brain that has visibility into all the things that are happening. You can see, for example, which teams and players are on which flights.

KALIOUBY: Wow, really?

HU: The reason this is so important is it used to be very static. You would get a report, and you’d have to sort through all these things, and you’d have to call someone else to find out the status. Historically, teams have had to hunt for data from other teams much more manually and then, A, put it together and, B, start to make it smart by correlating it. Here, you can actually see in real time how FIFA’s operations are advancing, and I saw examples. That team’s flight is late because we saw a thunderstorm, so we need to change the event schedule. But if we change the event schedule, we then have to get the operating officer there at a different time. You can start to see now, because everything is connected to everything else, that they can correlate things from flights to traffic patterns to schedules of leaders who need to show up, to where the coach is going to be, to the media events, so they can create a much smoother experience.

KALIOUBY: That’s so cool. It’s so great. I’m envisioning the command center. How many people work there?

HU: Oh, it’s hundreds, and I think the whole point is all those things that go into, “Hey, how did this World Cup feel?” Not just for the fans, but for the people who are participating: the referees, the analysts, the players themselves. All these are aspects where the intelligence behind the scenes is going to help make it much better for them versus what they’re used to.

Copy LinkThe command center keeping the tournament running smoothly

KALIOUBY: Yeah. So let’s talk about the fan experience. What is it going to feel like for fans who are lucky enough to be in the stadium live?

HU: Yeah. I didn’t mention this yet, but I think one of the big frustrations or challenges is just getting to where you need to go on time. We’re expecting, with FIFA, more than seven million fans across the 104 matches. One of the things that we’ve built with them is smart wayfinding, meaning, “How do you get to your seat? How do you get to the closest restroom?” So much changes about the event. If you think about all the things that change, traffic flows change, public transit options change, the flow within the stadium changes. I was actually stuck outside a stadium because it said, “Go to Gate 14,” but during the game, Gate 14 is closed. You actually have to go somewhere else. A normal app doesn’t know that.

We’re able to pull in the latest and most accurate, down-to-the-minute information relative to that specific match in that city on that day. That will help, and I think that’s going to be huge because that’s something we’ve heard a lot of feedback about as being challenging. That’s one way we’re using technology, real-time data, better telemetry, and greater accuracy, plus the integration of all that data, to make it meaningful for fans.

KALIOUBY: Yeah. Was there anything you thought of doing for the World Cup that you decided not to move forward with?

HU: I think it’s more a matter of degree in terms of how much autonomy or automation there is. It goes back to the question of when we think the technology might be doing too much. An example is, “Hey, you know what? We could make the referee cam super smooth. We can make the referee look like they’re just gliding on ice.” But then I think it loses something about being in the moment and the excitement and the reality of what it’s like at ground level on the pitch. So that’s an example where we said, “Oh yeah, we could make it look like it’s just on rails, but we want to make sure to keep the spotlight on the human aspect, the realism, and the dynamism of the matches.”

KALIOUBY: Yeah, I find that very interesting. You want to keep some of the messiness of the game, right?

HU: Yeah. And again, when you say messiness, it’s just another way of saying, “We want to keep the human aspect.” That’s why we’re not paying to watch robots with robot referees and robot coaches.

KALIOUBY: Yeah, I love that. I think what makes sports so awesome is the human aspect of it. It’s quite unpredictable, right?

HU: Yeah, so the guiding principle that we would use is very much around amplifying the humans. I think football is so dynamic that we’re never going to get away from having human decisions in the moment: “Do I make a pass? Will I shoot? How do I defend?” The art of the possible on the frontier is finding the critical moments that matter and applying AI to augment them. “How do we create more opportunities to score? How do I prevent myself from being on the back foot?” The decision-making framework all goes back to: Can we help put the spotlight on the people on the pitch who are playing the game so they can rise to their highest possible level? And that includes the preparation and the training, as well as the strategizing and the tactical execution. How can we help them focus on the game and less on the things that are distracting and not contributing to the things they really value and the passion of the moment?

KALIOUBY: Yeah. I love that as our ending thought: AI that helps unlock human potential. So, Art, thank you so much for joining us on the show, and I’ll be watching for all the behind-the-scenes heroes for this World Cup. Thank you.

HU: Such a pleasure, Rana. Really insightful questions, and I’ve enjoyed our conversation.

KALIOUBY: There’s so much happening behind the scenes to pull off one of the world’s largest events, like the 2026 World Cup. And the thing that stuck with me most in this conversation is this whole command center in Miami. Now I’m envisioning big screens with real-time analytics and data flowing everywhere, and I don’t know, I just think it’s so cool.

Are you watching the World Cup? Who are you rooting for? And have you noticed the presence of AI? If you have any thoughts or questions, I would love to hear from you. Thank you so much for listening. We’ll be back next week with a new episode.