Will AI Redefine Sports Judging?

On December 21st, 2024, Oleksandr Usyk, the WBC, WBO, IBF, and WBA heavyweight champion, defeated Britain’s Tyson Fury, for the second time, in a rematch from their February 2024 bout also held in the Kingdom Arena in Riyadh, Saudi Arabia. The fight made headlines because it was the first high-profile one also scored by an AI judge, which proclaimed that it has been put in place to ensure fairness. However, this judge was not an official one, and his scorecard did not affect the outcome of the fight. Per Saudi billionaire Turki Alalshikh, who promoted and organized the boxing match, this judge was only there to offer an unbiased, stat-driven perspective of the contest. It gave a final score of 118 to 112 for Usyk, which was a different scorecard than what the three official human judges sitting ringside gave the bout, as all three also scored it in Usyk’s favor, giving him the win, awarding him eight rounds to four, for a scorecard of 116 to 112.

The AI judge caused sizeable controversy, not only because it had a different score than its three human counterparts but because it scored two rounds, 10-10 or draws. It tracked landed punches, successful defense, and aggression, collecting real-time metrics to discover which opponent wins a round and by what margin. After its scorecard got revealed, many called it unimpressive, and that while they understand a desire for fairer results, technology is not at the point where it can score fights correctly, no matter what some may think.

That said, AI judging is already getting implemented in various sports, and it seems to be gaining traction, and may be on track to replacing human opinion regarding assessing athletic performance in some arenas. Naturally, this will affect sports betting, as gamblers will undoubtdley feel more comfortable regarding contest outcomes, their fairness.

Boxing & AI Judging

Even though Uysk-Fury 2 shed light on AI boxing judging and threw this topic into the mainstream sports discourse, this pairing is not something new. OpenAI, the creators of ChatGPT have created a system called Boxing Score, developed with one of the sport’s main organizations, the World Boxing Council, the WBC, to analyze boxing performance. The International Boxing Association (AIBA), also has its scoring system that uses AI-driven tech called Fight Score, doing pretty much the same thing as Boxing Score.

Then there is DeepStrike, an AI stat and judging system created for boxing, with versions of this software coming out for Muay Thai, BJJ, kickboxing, and MMA judging. It uses Jabbr Cam, named after the startup behind this technology. There is a minimum requirement of three of these cameras that sync with the Jabbr platform, which utilizes the proprietary DeepStrike AI that scans video footage to return automated analytics.

AI Judging Gymnastics

Yes, this is a thing, and it has happened via the introduction of JSS or the Judging Support System. That is a technology developed by Fujitsu with the help of FIG, or the International Gymnastics Federation. It is a system that utilizes HD cameras to capture movements. It then creates a 3D model of the athletes and their performance in a 3D environment for more precise analysis. It does this so that it can detect movements at angles invisible to our eyes. The hope here is that this will go a long way in minimizing scoring discrepancies, as judges can review routines like never before. We say this because JSS can analyze over two thousand elements with a degree of accuracy of 90%.

The Judging Support System, which started development in 2017, made its debut in Antwerp, Belgium, at the 2023 World Championship, where it got utilized to analyze the routine of Croatian gymnast Tim Srbic, whose Olympic qualification and medal rested on a challenge where JSS got employed. After analysis using the discussed software, the Croat’s score jumped by .2 points, which landed him a silver medal at the Worlds.

For many, this showed the value of AI judging in gymnastics and how it can be used to lower what in the sports gets referred to as leotard bias or when competitors from less high-profile countries get weaker scores on account of judges’ unconscious biases. Yet, some have complained that, while yes AI does offer objective analysis, it cannot factor in some of the most subjective elements of the sport, like the artistry involved and a performance’s unique qualities. Judgments to the quality of these rest on subjective judgment built upon years of experience.  

Judging Diving Using AI

Before the 2021 Tokyo Olympics, an AI Diving Judge project got presented by researchers at Brown University that put to use an Action Quality Assessment model that aims to accurately grade the execution and difficulty levels of dies. It does this through a multi-task learning model, using close to four hundred London 2012 Olympics videos and their scoring datasets, along with 3D video processing, like the gymnastic one discussed above. However, its creators admitted to this system’s shortcomings, and they have made no bones that its deficiencies lay primarily in its handling of different scenarios and low dataset diversity.

The problem with human diving judging is that it is inherently subjective, and judges do not have to explain how they got to their scores. This system, and the ones that followed after it, analyzed object detection and pose estimation. The software tries to extract a dive’s main elements, meaning movement on a platform and its height, splash quality, and a diver’s pose, using these to identify the dive type and segment its phases. Then, based on this, the AI runs a series of error analysis procedures to try and find faults, such as bad feet positioning, too much twist tightness, poor rotation accuracy, and so on. Each detected error then gets combined to produce an overall score. However, while functional, these systems have not been adopted at official competitions.

Will AI Judges Be Sports Mainstays?

Probably not in the next decade, as the technology is not yet available for most sports, and even in those where it is close, most believe that highly trained professionals can do a better job than AI, as proven in boxing tests. Still, hybrid models where AI suggestions get taken into consideration, like the gymnastic example, are likely to become common in multiple sporting arenas in the coming years, as organizers will probably believe that these will help boost fairness and judging accuracy. That should put bettors more at ease since they will no longer have to fear potential corruption as much.

Top Bitcoin Betting Sites