How AI will transform the Olympics

Artificial intelligence will rapidly change the Olympics in the years ahead, from identifying and training promising athletes with the help of machine learning to minimizing bias in judging. Whole cities will even use AI to host the Games more sustainably and with better outcomes for local residents.

That’s the vision the International Olympic Committee revealed when they announced their first-ever AI agenda ahead of the Paris 2024 Games. While few of these transformations may be visible this summer, University of Florida experts say the AI revolution will soon bring some of the biggest changes the international competition has seen in its 128 year history.

Perfecting talent

Underneath the camaraderie and friendly competition, the Olympics brings out a fierce struggle between nations for bragging rights. Countries pour millions into identifying and training the best talent in hopes of securing more medals than their chief rival.

Increasingly, countries have realized the benefits of finding top candidates at a young age to better hone their skills over the years. Now, AI is bringing some of the tech once only found in professional locker rooms down to the youth level where Olympics training starts.

“With wearable technology increasing and more youth athletes getting these technologies, the previous problem of not having enough data is rapidly going away. The new problems to solve are not having enough high quality data and needing additional power to analyze it,” said Garrett Beatty, Ph.D., a professor of applied physiology and kinesiology in the College of Health and Human Performance.

Identifying promising young athletes is just the beginning.

Training regimens are also being overhauled by AI. In April, the IOC promoted a table tennis robot trained on a gold medalist. Louisiana State University’s Jayden Daniels won the 2023 Heisman Trophy after a season spent training with hyper realistic virtual reality software out of Germany.

These machine opponents are unlike anything athletes have worked with before. Daniel Ferris, Ph.D., a professor of biomedical engineering in the Herbert Wertheim College of Engineering, has shown that our brains react differently when playing against robots than when playing against human opponents.

“It’s different, but AI could be better than training against a human opponent,” Ferris said. “If you can figure out how to put Lionel Messi’s leg on a robot, and you’re a goalkeeper who’s going to face Messi, getting 1,000 Messi shots from a robot kicker is going to be helpful, right?”

Achieving those outcomes will require more collaboration and organization, though. Right now, sports AI applications are in a bit of a “Wild West” phase, said Ferris, who is part of the $2.5 million UF and Sport Collaborative, funded by UF President Ben Sasse’s strategic funding initiative.

“There’s not a lot of structure, not a lot of data, not a lot of rules about how to interpret that data,” he said.

One of the limitations is secrecy. Private companies are scrambling to develop hardware and algorithms to sell to the highest bidder – usually professional sports teams. But they rarely share their underlying data, slowing down the development of the field as a whole.

And a lot of talent development is locked behind the secret programs guarded closely by individual coaches or teams. Beatty hopes that AI could improve access to world-class coaching.

“If we could start developing that talent in a more scientifically consistent way, it could have a major impact, and I think AI has the ability to help democratize that,” he said.

As Olympians are increasingly assisted by AI-powered training regimens, the Games’ judges are likely worried about job security.

Judging upgrades

Machine judging is perhaps most visible in professional baseball. Although umpires still call balls and strikes – for now – the human element is increasingly overshadowed by accurate measurements of strikes, ball velocity, rotation speed, and pitch style, all in a fraction of a second.

This machine-learning revolution in refereeing is only accelerating.

“I think you’d be hard pressed to find anyone who doesn’t say that in five or ten years, AI judging will be at least as good as a human is, but more consistent and accurate than a human is,” Beatty said.

Beatty is an expert in the effect emotions have on athletic performance. He has shown that athletes’ emotions can influence their reaction time and power. But it’s not just the athletes that have bad days.

“Judgment calls by officials can be highly dependent on emotion,” Beatty said. Travel and fatigue, life itself, make it hard to be consistent. Organizers want to turn to machines to control costs, reduce bias and avoid corruption.

But can AI really judge a diver’s poise, or the grace of a gymnast’s routine?

Even artistic sports, like diving, involve cold, hard math. At the launch of their AI agenda, the IOC demonstrated AI-powered analysis of tandem divers that can quantify body rotations and angles and the alignment between team members.

“But then what differentiates the biomechanical analysis of these performers with the art? That’s an answer that people have to come up with,” Beatty said.

“I think we’re going to settle on something where the quantifiable parts – that we’ve been bad at as humans – are handled by a machine, but we retain a human element for judging the artistic component,” he said.

Streamlining for host cities

Host cities typically have seven years to prepare to welcome over 10,000 athletes and millions of visitors. The time vanishes in a flash, said Kyriaki Kaplanidou, Ph.D., a professor of sport management in the College of Health and Human Performance.

She would know. Kaplanidou worked in the international media department for the Athens 2004 Olympics in her home country, where she saw firsthand the hurdles that host cities must overcome to pull off one of the world’s largest international events.

Her research has documented what she calls the “emotional cycle” of local residents’ response to the Games. First comes excitement, followed by skepticism and confusion as preparation for the  event starts to affect their home city. Then as the Games approach, excitement and pride return.

“You can potentially use AI to map out that emotional cycle and provide information to residents that helps them make the best of these unusual circumstances,” Kaplanidou said. “We could better alleviate their skepticism or further nurture their enthusiasm with improved communication.”

One of the largest barriers to success is the sheer amount of knowledge transfer that must be conducted for host cities to pull off an event of this magnitude. Over seven years, cities build up a small army of committee members who organize the competitions, venues, transportation, communication and security around the games. Few of those staffers have previous experience preparing for the Olympics.

Kaplanidou envisions using AI to maintain continuity between the games, so that previous host cities can share their wisdom with new ones. An AI module in each functional area for the Organizing Committee for the Olympic Games – such as the media area Kaplanidou worked in – could analyze when milestones need to be completed, the planning time and personnel needed, and how these goals link together optimally.

“There are companies already offering visualizations of the whole planning process within venue management,” Kaplanidou said. For example, the Paris 2024 Olympics used the company OnePlan to effectively manage some of their venue operations, using the kind of AI-powered digital twin software that will only grow more commonplace in the years to come.

Predicting the future is hard, these experts admit.

But as AI continues to transform daily life, we may see these facets of the Olympics – and others we can’t yet imagine – overhauled by machine learning by the time the Games return to the U.S. for the Los Angeles 2028 Olympics.