DeepMind and Liverpool FC develop AI to advise on football tactics

- Advertisement -


Corner kicks, like this one taken by Trent Alexander-Arnold for Liverpool, can lead to goal-scoring opportunities

Robbie Jay Barratt/ AMA/Getty

An artificial intelligence model can predict the outcome of corner kicks in football matches and help coaches design tactics that increase or decrease the probability of a player taking a shot at goal.

Petar Veličković at Google DeepMind and his colleagues developed the tool, called TacticAI, as part of a three-year research collaboration with Liverpool Football Club.

Read more


The future of AI: The 5 possible scenarios, from utopia to extinction

Corner kicks are awarded when the ball goes out of play over the goal line, and can be a good scoring opportunity for the attacking team. Because of this, football coaches develop detailed plans for various scenarios, which players learn ahead of games.

TacticAI was trained on data from 7176 corner kicks in England’s 2020 to 2021 Premier League season, including each player’s position over time and their height and weight. It learned to predict which player would be the first to touch the ball after the corner kick was taken. In tests, the receiver of the ball was among TacticAI’s top three candidates 78 per cent of the time.

Coaches can use the AI to generate tactics for attacking or defending corners that maximise or minimise the chance of a certain player receiving the ball, and of a team being able to take a shot at goal. It does this by mining real examples of corner kicks for similar patterns, then offering suggestions for how to change the tactics to achieve the desired outcome.

Sign up to our The Daily newsletter

The latest science news delivered to your inbox, every day.

Sign up to newsletter

In a blind test, football experts from Liverpool FC were unable to distinguish AI-generated tactics from human-designed tactics, and they favoured the AI-generated tactics 90 per cent of the time.

But despite its ability, Veličković says TacticAI is in no way intended to put human coaches out of work. “We are strongly in support of AI systems that amplify human capabilities and leave them more time for the creative part of their work, rather than a system that would replace them,” he says.

Veličković says the research also has wider applications beyond sport. “If we can model the game of football, we can model several aspects of human psychology better,” he says. “AIs, as they get more capable, they’re going to need to have a better understanding of the world, especially under uncertainty. Our system is capable of giving decisions and proposals under uncertainty. These are skills that we believe will be transferable to future AI systems, so it’s a good proving ground.”

Journal reference

Nature DOI: 10.1038/s41467-024-45965-x

Topics:

  • sport/
  • AI
FacebookTwitterEmailLinkedInPinterestWhatsAppTumblrCopy LinkTelegramRedditMessageShare
- Advertisement -
FacebookTwitterEmailLinkedInPinterestWhatsAppTumblrCopy LinkTelegramRedditMessageShare
error: Content is protected !!
Exit mobile version