Performance improvement, player monitoring, refereeing, comments… How artificial intelligence is flooding the world of sport

An artificial intelligence (AI) predicted that Ireland would emerge victorious in the Rugby World Cup, with France on its heels… If its clairvoyant gifts need to be reviewed, AI is already proving itself in other aspects of sport. From Ligue 1 clubs to the French XV and tennis, franceinfo: sport deciphers the advances in this technology.

In sport, the use of statistics and data (“data” in English) is not new. Finding its origins in the world of baseball, developing in the 1990s, particularly with video games, data has taken an essential place in sport since the beginning of the 2000s. Talking numbers has become part of sports jargon, and there is only a small step between data analysis and artificial intelligence. You still have to know how to take it and understand it.

Data, AI, men in the shadows

Fabien Galthié, coach of the XV of France, did it a little over three years ago. After meeting the director of the SAS Viya company at a seminar, he called on this company specializing in artificial intelligence to create a platform adapted to the needs of the French Rugby Federation. “We wanted to answer simple questions, he explained in a press conference in November 2022. How to train better to be better in matches? How are we going to beat our next opponent? What did the teams that beat him do and what did the teams that didn’t do?“

Sahbi Chaieb, data scientist, is at the heart of this collaboration between high technology and rugby. For him, the difference between data and artificial intelligence lies in the ability of AI to “be based on a large volume of data so that the machine learns and can make decisions or predict events. It refers to predictive analytics, used to anticipate the future based on patterns established from the past. Thanks to that interactive AI platform, the Blues staff can question with the aim of “oreject one’s intuitions”explains Sahbi Chaieb to franceinfo: sport. “The coach can ask : ‘What’s the best option for a camp outing?’ And there, the AI ​​will do a whole analysis to try to render in the clearest way the risks and opportunities of each option.”

In the service of performance

This predictive analysis is not divination. Rather, it is the logical continuation of what we call explorative analysis, which consists of grouping and classifying data to extract patterns. Nicolas Buffa, video analyst for the French rugby players, summarizes for 20 minutes his action as the will “to extract from this mass of data one or two essential pieces of information to share with the coaches, who will then establish the game plan and discuss it with the players.”

And it pays! A textbook case is that of the triumph of the Blues against the All Blacks in November 2021 (40-25). Using the platform, the Tricolores simply asked the AI ​​what differences existed between the victorious Black teams and the losing ones. Buffa explained in October 2022 in the Crunch podcast of The Team that they finally “managed to release information which was decisive in winning this match. It was data on the duration of possession of the ball by the opponent.” Clearly, those who lost to the New Zealanders exposed themselves by keeping the ball too long. A maximum time of ball possession and a maximum number of rucks has therefore been determined for the XV of France.

This effective tool is also being implemented in the football landscape. For two seasons, the Professional Football League (LFP) has implemented a system of tracking video in Ligue 1 and Ligue 2, transcribed by artificial intelligence. Through these videos, combined with the players’ GPS data, all the crucial information for the staff appears: average speed, peak speed, distance covered, decisive actions, passes… AI represents real added value: “Pto process two million lines [les données d’un seul match]it would take hours and hours of work, estimates Aurélien Dubearn, director of video analysis at Toulouse FC. VSsaved time, and this time can be reinvested elsewhere“. Then, it’s up to everyone to make the best use of the LFP data to get the most out of it.

“It’s like a cook. You have the ingredients, but you don’t necessarily cook the same way […] There is no a miraculous recipe.”

Aurélien Dubearn, director of video analysis at Toulouse FC

at franceinfo: sport

There remains a question of permanent concern: will AI, which can work day and night, replace men on the sidelines? Aurélien Dubearn concedes that at the speed at which it is progressing, it will take a special place, but there is no question of replacing the human factor: “Obviously, there will be a department dedicated to artificial intelligence in the teams, just as there is one for video, which did not exist ten years ago.” He also recalls that “football [et le sport en général], it remains human beings, with all the emotions and what that engenders. Will artificial intelligence be able to sense all the emotions in a locker room? It’s not tomorrow the day before.”

Filter, target information, evaluate, compare in a few clicks the individual performances of players and the results of a strategy… Ultimately, the AI ​​provides a quantified, precise and reliable report, which completes the experience of the staff. For Damien Comolli, the president of the Toulouse club, these methods aim to “remove as much as possible the uncertainties and irrationality that exist in football“, he said in The Team.

The Toulouse club is avant-garde in its use of new technologies. From 2020, it based its recruitment on an AI algorithm to identify talent. The tool developed makes it possible to condense all the statistics concerning a player and to propose the one which would be most adapted to what the coach wants. “With our budget, which is not one of the biggest in L1, we have to try to find solutions to be more efficient“, explained Aurélien Dubearn to France Football.

Referee, commentator… artificial intelligence is coming to stadiums

AI can also be useful for spectators. In view of the recent protests by French and then English supporters in the face of theisions of referee Ben O’Keeffe during the Rugby World Cup, refereeing remains more than ever at the heart of the debates. An artificial intelligence was able to count the errors made by O’Keeffe, reports West France. According to her, he committed 10, including eight which would have put the French at a disadvantage against South Africa. Can AI become a support for referees in the future? Even compensate for the lack of official referees in the lower divisions, as requested by ZDNet, a site specializing in new technologies?

Pierre Miralles, head of AI at Footovision, a start-up specializing in statistical analysis, also points to help in detecting match-fixing. “QWhen we look for players who perform, we can also find those who underperform.”he details for The Team.

Thanks to the power of artificial intelligence, in football stadiums, 3D images can be projected onto screens to define whether a player was offside or not. For Guy Accoceberry, former player of the French XV, the same model would be desirable in rugby, where the rules are sometimes unknown or poorly understood.

Towards a relegation of humans?

The ChatGPT plebiscite illustrates this; generative AI, which can construct in real time an oral or written response to a given situation, has made considerable progress. In July, this revolution found a direct translation: the comments on the summaries of certain Wimbledon matches were produced by AI. According to the organizers of the Grand Slam of tennis, the long-term objective would be for them to be able to comment on entire matches in tournaments which do not always benefit from commentary.

“Replace John McEnroe’s commentsit’s impossible !“, nuance Bill Jinks, the technology director of the prestigious London tournament.

“The human element must always be present, it’s simply a matter of complementing it. The challenge is to find the right balance between tradition and innovation.”

Bill Jinks, Wimbledon Chief Technology Officer

to AFP

Neither omnipotent nor ready to play the matches in place of the actors, artificial intelligence and its progressive democratization raise two questions: how far will it be able to go, and will it lead to a standardization of playing styles?