AI in sports judging is already here, bringing both advantages and potential pitfalls
Technology not currently being used to judge major competitions where subjectivity plays a role
Of the four judges used to score the boxing match between Oleksandr Usyk and Tyson Fury last December, one was missing a significant trait that the other three shared: Flesh.
In what was believed to be a first, AI technology was used, unofficially, to monitor the fight and score the bout.
While the AI judge agreed with its three human counterparts who gave the match to Usyk, it felt he had taken more rounds and won by a wider margin.
Although the AI scorecard was not part of the official results, its use in the bout was just the latest example of how the technology could start being incorporated in sports like figure skating and diving, where the subjectivity of the judges plays a major role.
While experts say AI may improve on certain issues that can hamper human judging, they suggest it can also present its own challenges.
AI may do a decent job judging the difficulty of a performance, says Willem Standaert, associate professor at HEC Liège in Belgium, where he researches sports digitalization. "Whereas the execution is where it might have more difficulty."
This week AI is being used to unofficially judge snowboarders competing in the halfpipe event during the X Games at Buttermilk Mountain in Aspen, Colo.
Video cameras will capture the moves of each athlete, and artificial intelligence will take that information and provide a score for the routine, The New York Times reported.
The AI at the X Games won't have any impact on the official scoring, but it will be a gauge of what's possible in the future
'Humans can get it wrong'
"Part of subjective sports, we see it all over the place, is that even at their best, humans can get it wrong," X Games CEO Jeremy Bloom told The Associated Press.
"Sometimes getting it wrong has huge implications. What if we could give judges superpowers and they could see things they couldn't see with the human eye, and this technology could help inform them?"
Some sports fans may already by accustomed to the use of AI as judges in sports where artistic merit has no bearing on the results of a competition.
At tennis's U.S. Open, for example, there have been no line judges — only chair umpires — since 2021, with Hawk-Eye Live electronic line-calling being used on all courts. Last October, Wimbledon organizers announced they would be replacing their line judges with electronic line-calling at the oldest Grand Slam tennis tournament.
But now, the technology is being considered in other sports where judging the difficulty and execution of the athlete's routine is part of the competition.
Last year, Colin Smith, the director of the International Skating Union (ISU), said the body was considering the use of AI to judge technical elements of skating, The Scotsman reported.
And in the release of the AI agenda for the International Olympic Committee last year, IOC President Thomas Bach said that AI "can help to reduce human bias in judging and refereeing."
Meanwhile, it's already being used for judge artistic gymnastics, but only in a supportive role. The technology, known as the Judging Support System (JSS), was first used at the Artistic Gymnastics World Championships in Antwerp, Belgium in 2023 for judging all apparatuses: vault, uneven bars, beam and floor.
Coaches can ask the judge to use the AI system to review a performance if they don't agree with the score.
According to the website of Fujitsu, the maker of JSS, the technology captures 3D data of the athlete's performance.
Movements such as joint angles, part of the judging scorecard, are captured and analyzed as numerical data, "and displayed in the same manner that judges need to see," according to the site.
Still, the technology has yet to be used as a replacement for judging in major competitions where subjectivity plays a role.
When it comes to judging the artistic merit of a routine, or its "flair," AI can run into problems, Standaert said.
"For instance, specifically on the floor exercise in gymnastics, there's music. How well does what the gymnast does on the floor align with the music being played? And so how are you going to get AI to judge that?"
But AI does have advantages over human judges, Standaert says, when it comes to certain potential biases, like those that may be based on an athlete's nationality.
He says it also has an advantage over a human judge's physical limitations — like poor eyesight, cognitive speed and just tiring after a long day of judging.
Resolve some biases, introduce new ones
But Standaert says their research with gymnasts has uncovered biases within AI, including a bias against larger, taller gymnasts.
Another issue with AI is skin colour, he says, noting that athletes or gymnasts with darker skin tones are apparently not able to be captured on video as well as athletes with lighter skin.
"By resolving biases you introduce new ones," he said.
Also, he suggested that while judges may reward an athlete's innovation in a routine, AI might say, " 'Well, this is not in the rule book. I don't know what to do.' "
For now, Standaert believes AI will continue to play a supportive role, and that for the artistic element of these types of sports, "they will have to keep a human in the loop for sure."
For his part, Bloom, the X Games CEO, says he doesn't envision a future without human judges.
"I don't think this replaces the judges," Bloom said, "but I think it gives them power to ensure objectivity."
ABOUT THE AUTHOR
With files from The Associated Press