It is the trail-blazing company that has conquered chess, gaming, and was this week hailed for “solving” one of science’s biggest riddles. They are the boffins behind the end of Liverpool’s 30-year title drought.
Now, artificial intelligence specialists DeepMind and the Premier League champions’ in-house research team have joined forces in a move that could lead to football clubs’ selection, tactics, transfer policy, and even training techniques, all being dictated by machines.
DeepMind and Liverpool have recently co-authored research on the potential use of AI in the game, which states their “overlying goal” would be the development of just such technology.
The research, entitled ‘Game Plan: What AI can do for Football, and What Football can do for AI’, envisages the creation of an Automated Video Assistant Coach (AVAC) to analyse match footage and “accordingly advise coaching staff” – including while a game is still on-going.
If the prospect of that being introduced into a sport still struggling to get to grips with Video Assistant Referees seems like the stuff of science fiction, DeepMind has form when it comes to creating AI capable of mastering the most complex of fields.
That includes the development of AlphaZero, which obliterated the previous highest-rated chess programme after taking just four hours to learn the game, and AlphaStar, which has attained Grandmaster status in the video game Starcraft II, regarded as an even bigger challenge.
But it is AlphaFold that has proven to be the Google offshoot’s most celebrated achievement after it was hailed this week for solving a scientific problem that had stumped researchers for half a century by showing it could predict how proteins fold into 3D shapes.
Liverpool's use of data was credited with helping them win their first top-flight title in 30 years
The breakthrough, which could help pave the way for designer medicines and the end of plastic pollution, was hailed by Venki Ramakrishnan, the president of the Royal Society, as “a stunning advance” that had occurred “decades before many people in the field would have predicted”.
The credentials of Liverpool’s contributors to the research into AI and football are almost as impressive.
Ian Graham, William Spearman, Tim Waskett and Dafydd Steele are widely regarded as the best team of their kind in the English game following the record-breaking run that saw the Anfield club crowned champions of England for the first time since 1990.
Graham, Liverpool’s director of research, has a PhD in theoretical physics from Cambridge University, Spearman is a Harvard University graduate who worked at the European Organisation for Nuclear Research, Waskett is an astrophysicist who specialises in coding, and Steele is a chess champion.
The quartet were already at the cutting edge of the use of AI within the game and are now looking to boldly go where no-one has gone before.
The research paper envisages the creation of AVACs through a combination of what are called computer vision – the automatic extraction and analysis of useful information from digital images or videos; statistical learning – a mathematical machine-based framework; and game theory – the study of how and why people make decisions.
“A successful AVAC would help players, coaches, and spectators alike,” the authors write. “Specifically, it could help the players by analysing their play for weak points to further develop. A player’s performance throughout a game could be analysed to suggest improvements in position play and assessing the performance overall.
“Prior to a game, an AVAC could suggest strategies tuned to the opponents of the day. Coaches also seek to get the best out of their players, but have a limited amount of time and many players to observe and provide feedback to.
“An AVAC would offer coaches many opportunities to help individual players and the team as a whole, suggesting player rosters for a given game, as well as trading or scouting strategies based on counterfactual evaluation of team performance with brand new players.”
There could also be benefits for spectators.
“Such an AVAC system would have the ability to automatically sift and label huge quantities of video streams, enabling broadcasters and spectators alike to retrieve key moments. An AVAC could automatically keep a running tally of information the spectator may find interesting based on their reaction and the current state of play.”
In one example of what AI methods could do, the researchers combined statistical analysis with game theory to devise penalty-kick strategies based on individual playing styles, including how they move on the field. In another example, they used computer vision and statistical analysis to compute the best predicted trajectories for players to take as they ran down the field and looked for situations in which a player’s actual movements deviated significantly from this predicted trajectory.
The authors concede “football is arguably the most challenging to analyse of all the major team sports” but nevertheless conclude: “Football analytics poses a key opportunity for AI research that impacts the real world. The balance of its reasonably well-controlled nature (versus other physical domains beyond sports, e.g., search-and-rescue), considerations associated with human factors (e.g., heterogeneous skill sets, physiological characteristics such as injury risks for players, etc.), and the long-term cause-and-effect feedback loop due to the relative infrequency of scoring even in professional play make it a uniquely challenging domain. Nonetheless, the rapidly-emerging availability of multi-modal sensory data make it an ideal platform for development and evaluation of key AI algorithms, particularly at the intersection of the aforementioned fields of statistical learning, computer vision, and game theory.
“Overall, the combination of data sources, downstream benefits on related domains, and potentials for impact that AI could have on the football domain are quite evident.”
Of course, all this raises the question that if a machine could select the best players, tactics and training techniques at a football club, where would that leave the manager?
Whatever the answer, as VAR has all-too evidently proven, the use of technology in the game is only heading in one direction – for good or for ill.