Technology: Kick with AI: How AI is transforming football

Researching players, improving training, tactics and match analysis: artificial intelligence has become an integral part of football. But does technology also have the potential to completely replace coaches?

A new AI application aims to improve corner kicks in the future. This is just one example of how artificial intelligence can change football. Different AI tools are already supporting sports in different areas, so could the technology make coaches unnecessary? A German expert reassures.

Artificial intelligence (AI) has already been used in various areas of professional football, such as recruiting new players. Many software companies have developed artificial intelligence systems that – quite simply – evaluate and then evaluate players' gaming and training data. If clubs are looking for the best possible addition to their team, AI systems provide them with suitable candidates from huge databases that will improve the team. In this way, the AI ​​results provide data-based decision-making aids for clubs' human scouts.

A countless amount of data is collected in professional football: body sensors produce performance data during training which allows statements to be made about the player's training status and development. High-resolution cameras document every movement and location of all players and the ball on the field. The information generated in this way not only supports live reporting, but also allows AI-powered analytics to evaluate game tactics, evaluate individual players and improve future tactics.

In 2021, the field of “Planning, Strategy and Optimization” was described in an expert report of the Federal Institute of Sport Sciences (BISp) as probably the most interesting sub-step in the application of AI in sport: the focus is “on using models to improve tactical behavior with regard to performance.” The game or the success of the game.” The project “Multimedia Analysis for Sports Analytics” (MM4SPA), funded by the Federal Ministry of Education and Scientific Research, aims in a similar direction and deals with the analysis of website and video data.

Better angles with artificial intelligence?

TacticAI, a system developed by the Google Deepmind team in collaboration with Liverpool Football Club, is specifically related to corner kick tactics. As the research group reported in the journal Nature Communications, the AI ​​was trained using data from nearly 7,200 corner kick situations that occurred in the Premier League between 2020 and 2023.

According to the study, and based on this data, “TacticAI” is able to provide information about how to position individual players in the corners so that the shot leads to the goal. According to the research team, the recommendations issued did not differ from the recommendations of human trainers. Additionally, in a poll, soccer experts preferred AI tactics over human recommendations in 90 percent of cases, according to the study.

In light of the study's findings, the question arises as to whether artificial intelligence could make human trainers completely unnecessary in the future. “It certainly won't come to that,” confirms Daniel Memmert, a specialist in sports information technology at the German Sports University in Cologne. The use of artificial intelligence in sports usually involves so-called machine learning approaches. Machine learning describes a subfield of artificial intelligence that deals with the development and application of statistical algorithms. Identifying problems that AI applications can help solve, choosing the right AI tool, and interpreting and weighting the solutions generated by AI—all tasks that must be performed by humans, says Memmert.

Moreover, AI is powerless without data. People are needed to make it available and process it. “In general, the process of using AI still involves a lot of human parts at crucial points,” Memmert summarizes.

“Tradition does not score goals.”

According to Memmert, it is “clear” that major clubs these days can no longer do without artificial intelligence systems, which is dissatisfying some fans who fear that automated support will take away the spirit of the game. “I can understand nostalgia, but tradition doesn't score goals,” Memmert says. In addition, anything new is often rejected at first. “We also see that through the video evidence, though, that she has made a lasting contribution to the fairness of the game.”

If coaches don't become redundant, maybe players will? However, Google's DeepMind group recently introduced small humanoid robots that, thanks to a special kind of machine learning, become surprisingly agile and dynamic soccer players. Videos from the study published in the journal Science Robotics show players confidently dribbling, quickly getting up after falling and quickly changing direction in surprisingly complex 1:1 games.

In the clips, in addition to the still somewhat clumsy action sequences, one difference becomes clear from the human games: the little robots don't celebrate the goal.

al, dba