Stones of the Yarra Valley

Growing figure skating controversy sparks big question: Can AI fix officiating in sports?

· Yahoo Sports

LIVIGNO, Italy — If you could snap your fingers and remove officiating mistakes in every sport, would we have the same Super Bowl winners, NCAA champions and Olympic gold medalists that show up in the history books?

Visit goldparty.lat for more information.

It’s an impossible question to answer. But it’s one former Olympic skier and football player Jeremy Bloom wishes we didn’t have to ask.

“Teams and individuals lose well-deserved winning moments because of human error,” Bloom told Yahoo Sports. “Being an athlete, understanding how hard it is to climb that mountain, I think everybody – literally everybody – should be united on a front of ‘we can’t make mistakes.’ These moments are too big. That’s the problem I think all of us that care about these athletes and these sports should be trying to solve.”

Owl AI, the company Bloom founded, might be part of the answer.

At the midway point of these Winter Olympics, we already have one judging controversy threatening to consume the discussion around figure skating. 

It involves ice dance, where a French judge’s scorecard showed a larger gap between a French team and American team than other judges in one of the components. The Americans, Madison Chock and Evan Bates, settled for second while the French team won gold.

Though there’s been no formal accusation of wrongdoing, the controversy has echoes of a scandal from the 2002 Games involving a French judge who allegedly agreed to boost the Russian team in pairs figure skating in exchange for judging help with a French team in a different event. 

At the beginning of these Olympics, there were also questions surrounding the score given to Chinese snowboarder Su Yiming, a former gold medalist, who did not perfectly execute his trick in Big Air but was scored high enough to knock American Ollie Martin off the podium. 

In many ways, these controversies are inherent to judged sports like figure skating, snowboarding and freestyle skiing. Subjectivity and unconscious biases come into play. The pressure on judges to deliver scores quickly can lead to mistakes. 

What if the answer to all that is artificial intelligence? Bloom, who has raised $11 million in seed funding for Owl AI, is on a mission to figure out what’s possible. And as the CEO of the X Games, Bloom is already putting the product to work on a limited scale with bigger plans for the future. 

“What we’ve found today is, it’s an incredible judge,” Bloom said. “It’s showing — it’s got to be proven — but it’s showing objectivity. I think we’re just continuing to throw everything at the technology to see where it’s good and where it’s not.”

It’s unclear, ultimately, how important AI will be — and how important humans want it to be — in officiating sports.

Professional tennis has already replaced line judges at most tournaments with a form of AI that instantly calls shots in or out. Some fans and players like the objective nature of the system; others don’t trust the technology to be 100-percent accurate and believe a layer of drama has been lost with players no longer having the ability to challenge calls they feel were incorrect.

The next level of possibility is more complex — and controversial. Imagine a world where you’re watching an NFL game and a computer immediately flashes a graphic on your television screen whether a pass interference penalty should be called. Or perhaps an NBA game where there’s no need for a coach’s challenge on a controversial block-charge call because AI instantly gives us the final word.

There have been growing questions about the juding in the ice dance competition that awarded the gold to Laurence Fournier Beaudry and Guillaume Cizeron of France (right) over Americans Madison Chock and Evan Bates. (Photo by Tim Clayton/Getty Images)Tim Clayton via Getty Images

Would it be a fairer system for the athletes? Probably. Would it be as enjoyable to watch without the controversy and human element? That’s in the eye of the beholder.

But the impact of AI isn’t going away, and in many ways Olympic sports are an ideal canvas for experimentation — even if some competitors have reservations about what it could mean down the road.

“Our sport, and judged sports in particular, there’s a level of artistry that I don’t think an AI could really judge — or at least that anybody would feel good about,” said Nick Goepper, a freestyle skier with medals from the last three Olympics. “There’s some intangible factors you have to put into play like, ‘Has this ever happened before? How does a new trick affect the sport and culture as a whole?’ There are some of those audibles that a human judge can throw when you really understand the larger scope of things and connect to the sport on an emotional level.”

At their most fundamental level, though, sports like snowboarding and freestyle skiing face a judging conundrum. Each year, competitors continue to advance and push boundaries, executing harder tricks with more mid-air rotations and subtle stylistic elements that can be difficult to pick up. A winning routine at one Olympics is likely to be considered passé by the next.

In a sport like big air, where competitors jump off a ramp and get scored from 0 to 100 on one trick, judges are expected to identify and score a variety of elements including amplitude, rotations, inversions, grabs and landing. And even though they have instant replay available, asking judges to deliver scores quickly on these complex tricks — usually within about 90 seconds to two minutes — is in some ways unfair to them, not to mention the competitors.

“It’s tough because a lot of the judges have never performed the tricks we have done,” freestyle skier Alex Ferreira said. “The level is so high they’ve really had to lock in and pay more attention. In a perfect world there would be more time. In the moment, the weight and the pressure is so heavy to get the score out that it can probably lead to some mistakes. But for the most part they’re doing the best they can. I would hate to be a judge.”

A properly trained AI could, theoretically, both identify all the technical elements of a trick and give some context to degree of difficulty without bias or giving benefit of the doubt to more famous competitors. Whatever unquantifiable judging advantage Shaun White might have had in the Olympic halfpipe simply by virtue of being Shaun White, that goes away when AI is making the call.

So far, Bloom has been blown away by the results.

“Our judges have been part of this process,” he said. “We had to teach it what good style looks like. That was a fun challenge, and it turns out good style is just good economy of motion in the air. Is the rider on axis or do they throw up a hand because they missed the take off and they need to get back on axis? What is a good landing and what is a great landing? What is a good grab and what is a great grab?”

For now, AI replacing a judging panel seems like a bridge too far. But under Bloom’s leadership, X Games has been integrating it into the experience since last year. 

At last month’s X Games Aspen, Owl AI was used not only to project scores as soon as riders completed their runs (the AI scores were not factored into the outcome this time), a human voice predicted winners based on the AI’s evaluation of practice runs and translated commentary into various languages for YouTube viewers around the world. 

Also, for the first time, judges were given the AI breakdown of what occurred during a trick to help them with their scores.

Jeremy Bloom has gone from the Olympics to the NFL to entrepreneur trying to eliminate human error in sport officiating. (Eugene Gologursky/Getty Images for Fast Company)Eugene Gologursky via Getty Images

“Providing real time superpowers to human judges is part of the strategy,” Bloom said. “Was that a tail grab or a mute grab? How many rotations was it? What was the amplitude — 12 or 13 feet. I think in a perfect world today, it sits amongst the humans for sure, not replaces humans.”

As these sports keep progressing, often faster than the evolution of the judging, it could become a necessary tool. But some athletes fear that a critical element of what makes their sports compelling will be lost if AI is allowed in the door. After all, AI is only as good as what it’s taught, which might stifle artistic expression if athletes are trained to perform for what the AI values and not a more malleable, emotional human experience.

“What is correct technique? There’s not necessarily one correct way,” figure skater Amber Glenn said. “It is an artistic sport. There’s always going to be an opinion.”

Here’s another issue: In many of these high-leverage competitions like the Olympics, competitors will debut something completely new that the sport has never seen before. That’s what won freestyle skier Alex Hall the gold medal four years ago in slopestyle when he executed a “double cork 1080 bring back,” which became known as the pretzel because it required him to stop his rotation mid-air and almost defy physics by pulling back the opposite direction.

In other words, while the raw number of rotations is often the separator in these events, Hall impressed the judges with creativity. He’s dubious about AI being able to account for that.

“It’s so niche and there’s an element of execution style that’s so subjective,” Hall said. “I’m not doubting it. I think it could work at some point. But I kind of like the human nature of it, and it’s slightly imperfect in a way. I know it’s not great for a competition but anyone who’s in freestyle skiing gets that and is OK with the chance of it not being perfect.”

There’s no way to predict where all this leads. Could we see a future Olympics where AI is utilized either as a tool to help judges or to provide some component of scoring? It’s far too soon to say.

But there’s no doubt it has potential to disrupt longstanding officiating and judging practices across an array of sports as the technology is refined and gradually implemented into events like the X Games.

“It'll never solve pass interference because that’s a subjective call, but if this technology can call it the same whether it’s the fourth quarter or the first quarter, whether it’s a superstar or someone you’ve never heard of and create a level of consistency around that call, that’s the goal and objective,” Bloom said. “Whether it’s a $5 billion or a $100 million (company) matters a lot less than us trying to figure out how we can make sports more fair so that nobody is sitting on the sidelines when they should be hoisting the trophy. It’s not an easy mission at all, but it’s an important mission.

Read full story at source