The 2026 World Cup has come and gone, and I'm sure many of us are feeling the gap it's left in its wake. Maybe I'm starting to realize a weird pattern in my thought processes (see this article) but one of the things that I really thought about during the event was about the types of AI being used in sports analytics and what it can tell us. Let's dive into it, because I did a little bit of research on this and need to get my thoughts out somehow.
FIFA used Lenovo during the world cup for its "AI-powered analytics and AI-enabled stabilization for on-field referee cameras, to 3D player avatars for improved semi-automated offside technology and tablets integrated into select fan sections for a more interactive, semi-immersive experience". Interestingly, it wasn't a completely smooth integration all around, and it took quite a few conversations with teams until "[the] teams [were] on board with the technology because they could clearly see how it would improve rulings on the field, decrease the delays associated with lengthy reviews, and improve analytics for team strategy."
With literally almost all the eyes in the world on the event because of its explosive popularity, Lenovo had the daunting task of making sure that the absolute minimum number of errors occurred - after all, FIFA is... a pretty HUGE client. Ultimately, here's what FIFA says is their goal with the way they're embedding AI into the sport:
"The ultimate goal: To create better competition by elevating every team's chance of winning through smart AI and data"
The specific AI being used is called Football AI Pro which is an AI knowledge assistant designed to support the 48 participating teams. The goal in providing every team with access to this technology is to minimize discrepancies in financial backing, since advanced data analysis and predictive tests can be extremely expensive.
This is what you can blame for all of the offsides during the tournament, with "a significant development in semi-automated offside technology" through the use of the 3D player avatars. Scanning to create the avatars takes one second per image, so the VAR has the most accurate frame-by-frame reference to somehow STILL make the wrong call. All jokes aside, the technology isn't actually new, it was already tested at the FIFA Intercontinental Cup.
Another cool feature of this AI is its Referee View, which apparently claims to have "footage captured from the referee’s camera will be smoothed in real time, reducing motion blur caused by rapid movement" to help with better insights into incidents.
It's an interesting and profoundly wide-scale deployment, and it has the potential to seep into becoming standardized (on an obviously much smaller model scale) in regular UEFA Champions League season matches. It's an interesting technology to be aware of, because now when you blame the refs for making the wrong call, you can cite these technologies at your disposal.