Artificial intelligence (AI) is becoming increasingly important in sport: it predicts performance, discovers talent and can prevent injuries. Will she soon replace coach?
Where there is a lot of data, artificial intelligence (AI) feels at home. The more the better. If she gets enough food, she can do amazing things: predict performance, recognize mistakes in posture or analyze movements with pinpoint accuracy. In sports, the use of AI opens up unimagined possibilities.
But not everyone in the scene shares their level of knowledge. Therefore, the question of how AI is already being used in sports is not so easy to answer. “Artificial intelligence can help to structure data, show abnormalities and reduce the amount of data so that people can deal with it better,” explains Carlo Dindorf, sports scientist at the Technical University of Kaiserslautern-Landau. But what does that mean in concrete terms?
YouTube video from SWR Sport: “Artificial intelligence in sport: Opportunity or risk? | SWR Sport”
Where is AI already being used?
Artificial intelligence is already able to predict the fitness of athletes. In performance diagnostics, vast amounts of data are collected that can be fed to the AI: training data, heart rate, oxygen and blood values. The TU Kaiserslautern has taken advantage of this: Over time, the AI learns to predict the results of performance diagnostics. Pretty practical, because the sports examinations are complex and expensive. Thanks to the AI, the athletes no longer have to torture themselves in the laboratory and always know how their performance is doing.
In performance diagnostics, an AI can predict the results based on training data.
In a smaller form, something like this has long been part of everyday sports life. Fitness trackers with information on stress and fatigue responses are standard. At the German cycling team Bora-Hansgrohe, six people are already working on the daily processing of the vast amounts of data on wattage, heart rate and much more. In Formula 1, data analysis has been influencing the racing strategy for many years, and the use of AI via software companies is also being intensified here.
When analyzing misalignments, AI can deliver results that a doctor can hardly keep up with. A scanner is used to create a 3D avatar of the athletes, which is then examined digitally. The results and patterns recognized by artificial intelligence can be used by trainers and therapists to prevent injuries and increase performance.
A 3D scanner uses AI to evaluate an athlete’s body.
Scouting in professional football
Werder Bremen became aware of goalkeeper Jiri Pavlenka through digital scouting. FC Bayern Munich also uses the possibilities of AI in transfer planning.
In professional football, despite initial successes, many clubs still have reservations about artificial intelligence. Despite the proximity to the Technical University, according to managing director Thomas Hengen, 1. FC Kaiserslautern has not yet dealt with it. Mainz coach Bo Svensson has not heard of the use of artificial intelligence in football either. Freiburg coach Christian Streich is concerned about the topic: “If you only used it for the positive things, it would be progress for mankind. But the negative factors are not foreseeable, it could hollow people out.”
What is possible in the future?
A current field of research at the Technical University of Kaiserslautern-Landau: Movement analysis. The athletes are observed during their exercises in a virtual reality environment: the artificial intelligence finds out where mistakes creep in and what could be done better.
Artificial intelligence (AI) is becoming increasingly important in sports.
This would make it possible to compare your own movements not only with yourself, but also with those of other athletes. This offers unimagined potential: You could learn training from the best in the world, learn their movements and find out where and why you are still lagging behind compared to the world’s best. In addition, the AI could recognize in the movement that an injury is imminent that can be averted at an early stage. All of this will be possible in the medium term, explains sports scientist Carlo Dindorf.
The Institute for Applied Training Science (IAT) in Leipzig sees it similarly, where AI is now used primarily in biomechanics. “In our area, AI is not a risk, but an opportunity to generate more data in a shorter time,” explains Björn Mäurer, scientific IAT sports informatics employee. In the IAT, the movement of the athlete on the ski jump or in the discus ring is filmed, the recordings are then evaluated with software and the athlete is accompanied by AI-supported recording systems. In an international comparison, Mäurer said they are “in a good position in this regard, but China should be further along”.
In professional football, the Plaier company wants to take player scouting to a new level using specially developed AI. A real-time analysis of the game system and squad is used. The results are combined with data on over 100,000 players registered in the system and weighted according to their skills in relation to the searching club. With this systematic approach, the AI learns from historical data and forecasts. Co-founder Jan Wendt promises customers a 90 percent probability of success in transfers: “We’re not saying: That could be. We’re saying: That’s the way it is and will look like this in the next six years.”
Ex-Schalke coach Manuel Baum, on the other hand, is working on an AI assistant coach to support the coaches during a football game. Through the live analysis of game data, the AI should provide information on possible substitutions and tactical adjustments. The speech during the half-time break could also be determined by the AI.
AI support in trampolining
At the federal base for trampoline jumping in Bad Kreuznach, data has been collected diligently for two years and fed into an AI. This could soon support the referees and coaches. Because trampoline jumps are highly complex, the rapid sequences of somersaults and twists are often not visible to the naked eye. The AI uses a sensor that the athletes wear on their bodies to learn how to recognize and analyze jumps.
The AI could support the judges in competitions to determine the level of difficulty of an exercise. The coaches would also benefit: Through detailed analyzes they could work specifically on weak points during training. The AI is not yet used in trampoline jumping. However, the research team estimates that it could be so far in one to two years.
Can AI replace humans?
Scientists largely agree. AI is far from being able to replace humans. Although she works faster, better and more reliably in parts, it is hardly possible to collect so much data that it comes close to the assessments of a sports supervisor. Should this be the case at some point, a whole new question will arise for sport: who is actually responsible for decisions made by an AI?
German Ethics Council on AI
The German Ethics Council positions itself in its statement “Humans and machines – challenges posed by artificial intelligence” for dealing with AI. These are the recommendations:
- AI must not replace humans
- AI should support decisions and not take them away
- Man should always exercise ultimate control
- It must be clear how the decision of an AI is made
- People’s interests should always be the focus
- Intrusions into privacy are to be prevented
Even most affected athletes cannot imagine a pure AI trainer. The junior world champion in trampoline gymnastics Aileen Rösler would like to be addressed personally during training: “The AI can be helpful to evaluate jumps, but the personal component, the address from a trainer, cannot be replaced by an AI.” Track cyclist Jule Märkl, who relies on AI-based performance diagnostics in Kaiserslautern, also knows: “If I don’t feel good, artificial intelligence can tell me what it wants, then I still don’t feel good.”
The potential of artificial intelligence in sports is huge. The way in which it will be used seems to be still open in many areas. The fact is: Sport will go into the future with AI.
