Google's AI arm, DeepMind, has developed a tool that can analyse expert football tactics and predict the outcome of corner kicks.
The Google-owned artificial intelligence (AI) company is no stranger to surpassing human game capabilities. It has previously conquered "Go," a complex board game and a challenging real-time strategy video game, "StarCraft II."
🎲 TacticAI uses both predictive and generative AI to help redesign corner kick tactics to optimize for certain scenarios.
— Google DeepMind (@GoogleDeepMind) March 19, 2024
It can also propose adjustments to players' positions - meaning coaches can identify patterns and detect who is critical to a tactic's success or failure. pic.twitter.com/AU3jfJ1N7Z
Cashing in on the capabilities of AI again, DeepMind unveiled TacticAI, a three-year project developed in collaboration with Liverpool Football Club, in a blog post by Zhe Wang and Petar Veličković on Tuesday. According to the blog post, TacticAI achieves "state-of-the-art results" by combining generative and predictive AI.
We're announcing TacticAI: an AI assistant capable of offering insights to football experts on corner kicks. ⚽
— Google DeepMind (@GoogleDeepMind) March 19, 2024
Developed with @LFC, it can help teams sample alternative player setups to evaluate possible outcomes, and achieves state-of-the-art results. 🧵… pic.twitter.com/8oVYtamGGe
"TacticAI demonstrates the potential of assistive AI techniques to revolutionise sports for players, coaches, and fans," DeepMind noted. TacticAI, a complete AI system, empowers coaches to explore various player formations for set pieces (like corner kicks). Next, the tool simulates the potential outcomes of these different strategies, helping coaches to evaluate their effectiveness directly.
🟩 In a blind test with football experts, TacticAI’s suggestions were indistinguishable from real corners in previous games.
— Google DeepMind (@GoogleDeepMind) March 19, 2024
When the group were asked to choose preferred tactical advice, they favored TacticAI's ideas over the original 90% of the time. https://t.co/XPPg2BgMJK pic.twitter.com/uzwBSbKfaj
TacticAI even impressed in a blind test, where its generated tactics were indistinguishable from those created by human experts. It is also worth noting that coaches favoured TacticAI's setup suggestions, choosing them 90 per cent of the time over human-designed tactics.
AI assistant or coach replacement?
DeepMind's TacticAI stands in contrast to Elon Musk's recent warnings about AI replacing jobs. DeepMind stresses that TacticAI is not a replacement for coaches. Instead, it is designed to assist them. This motive aligns more with the views of AI pioneer Professor Yann LeCun, who has cautioned that artificial intelligence may still require decades to achieve human-level understanding.
Good old-fashioned coaching may be replaced by AI assistant to analyze corner kicks?
— Sukumar Satyen (@sukumarsatyen) March 20, 2024
If that key to winning matches lies in crunching numbers & sampling player setups, then actual football strategy would rely on algos to kick to victory!
Then robots taking to pitch alongside.
"TacticAI's generative model also allows human coaches to redesign corner kick tactics to optimise probabilities of certain outcomes, such as reducing the probability of a shot attempt for a defensive setup," DeepMind said.
📊 Corner kicks can be challenging for AI to model due to the limited availability of data - @premierleague matches only average about 10 a game.
— Google DeepMind (@GoogleDeepMind) March 19, 2024
TacticAI uses a geometric deep learning approach to tackle this problem. → https://t.co/khfP2CseyV pic.twitter.com/SeOJMTenwj
TacticAI's analysis allows coaches to identify crucial patterns and key players that influence the effectiveness of a particular strategy without breaking a sweat. The AI tool, trained on data from multiple real-world corner kicks, helps coaches experiment with and analyse various player formations for set pieces.
DeepMind believes TacticAI could offer tactical advice in other sports with further development. Aside from this, it could shed some light on human psychology. This capability could be helpful in areas like robotics and traffic coordination. Notably, TacticAI will address three core questions:
Given a particular corner-kick formation, TacticAI can predict the play's most likely outcome. For example, it can identify which player is most likely to receive the initial pass and calculate the probability of a shot attempt.
Following a corner kick, TacticAI can analyse and compare the outcome to its prior predictions. This knowledge allows coaches to adjust their strategies for future set pieces by understanding how similar tactics have performed in past games.
Google DeepMind's new AI tool is not just applicable to predicting outcomes. It can enable coaches to explore adjustments that might impact the game's course. For instance, they can virtually reposition defenders and see how it affects a shot attempt.
