Sports have always had a human and data-driven element, but for the first time this summer, the management and enjoyment of the pinnacle of sports competition was aided by artificial intelligence.
When the 2024 Paris Olympics kicked off in early August, divers could view a 3D graphic of their time and speed with the help of a computer vision system. Athletes had a chatbot to answer questions and an AI-powered review system checked social media to ensure abusive comments aimed at them were taken down. Officials know in advance when they need more power or what arenas of the Olympic and Paralympic venues may have accessibility issues thanks to a "digital twin" of those places generated through a partnership with Intel.
With so much AI in play, the International Olympic Committee needed a plan to manage its own uses and explore how the technology could be deployed safely. It convened a working group months in advance to ask those questions, with the results being the Olympic AI Agenda.
Ilario Corna, the chief technology and information officer of the IOC, said during an 8 Aug. press conference the creation of the document helped ultimately put humans, not the technology, at the forefront of the Olympics.
"It may not always be that evident, while you are dazzled by the world records and extraordinary feats of the athletes," he said. "But in the background, technology is working towards changing the way we deliver the games."
The resulting document, participants of that group say, show some tenants of AI governance are universal, no matter the venue: consider the risks, seek different opinions and be transparent. Those lessons could be helpful guideposts for other sport federations if they seek to adopt AI as well, according to Deloitte Partner Daniel Ferrante.
"All these data can be connected in different ways, and maybe surfacing some of that connectivity can be helpful," Ferrante said, pointing to instances such as using data to determine if an athlete might be using performance-enhancing drugs without infringing on privacy or transparency.
He added, "I think those conversations are super important, because having that kind of framework from the very beginning and then potentially trickling down and helping to lead other federations is super relevant."
The agenda itself focuses on how to support athletes and keep the games safe, ensure equal access to AI's benefits, enhance the sustainability of the operations, create better engagement with fans and improve efficiency.
Those focuses can take different forms. For instance, the IOC says it believes AI can help identify talent by studying performance metrics, potentially lifting unrecognized athletes' profiles around the world. But to do that, the technology needs to be available to all, which could be done by educating others about its uses or making AI available to coaches to help create training plans.
It also notes AI usage in the Olympics should support equity and sustainability, be robust enough to remain reliable, its usage transparent and its underlying data and privacy policies aligned with public expectations, including limiting what information is collected and ensuring it is protected.
But the document also makes clear the IOC sees its role for now as one of highlighting AI's strengths and weaknesses and to lead by example, rather than being the Olympics’ overall AI governor. Creating a governance framework is a priority in the long run, the document notes, and the IOC is "committed to defining a path for AI that is both ethical and trustworthy.”
IOC President Thomas Bach during the launch of the agenda called for governments to pass AI governance laws to steer the technology's use.
Looking at AI from a high-level perspective was aided by having people from differing disciplines participate in advising on the AI agenda, University of Oxford Associate Professor of Operations Management Agni Orfanoudaki said. In addition to Deloitte members — the organization has a 10-year contract with the IOC to help create more efficient Games — there were representatives from the AI industry, athletes and academia.
It also helped the group define the IOC's role as a potential pioneer, facilitator and guardian of AI within the context of the Games, Orfanoudaki said. While she focused on the data available to the IOC and how it could be used to improve the audiences’ and athletes’ experiences, she said the process should be seen as a roadmap by other institutions about how to approach AI strategically, rather than dealing with it in a case-by-case basis.
"We wanted to form a roadmap for the IOC that is flexible enough but at the same time comprehensive," she said.
KU Leuven computer science professor Jesse Davis noted some risks are outside the IOC's control. Information such as health data is collected by a national federation or a sport’s governing body; a third-party application’s practices are unique to that company. Transparency into how an AI system works and how its trained in sports mirror greater-scope conversations around AI taking place all around the world, he noted.
And there are some risks unique to sport itself, Davis said. The analysis of basketball and baseball performance has fundamentally changed those industries, which can create a more homogeneous style of play and be a turn-off to fans.
In instances such as using cameras to determine if a soccer player was offsides while making a goal, Davis indicated technology can alter a fan's experience.
"It should be fun to watch a game, and some fans don’t want to wait for a (video-assisted referee) check to celebrate a goal," he said. "Moreover, there are some complaints about someone’s elbow making them offsides."
Chad Deweese, a senior manager in Deloitte's sports and strategy practice, said the AI Agenda also makes a point of saying AI should be a complement to the human experience of the games.
"A lot of it boiled down to, don't use AI because AI can do it," he said. "Use AI because we need something else to help, and that's the use case for AI, not just because AI can."
Caitlin Andrews is a staff writer covering AI governance for the IAPP.
