The partnership between Liverpool and Google DeepMind, initiated three years ago, has yielded a “full AI system” known as TacticAI. This innovative tool employs predictive and generative AI techniques to analyze a comprehensive dataset comprising 7,176 previous Liverpool corner kicks. By leveraging this data, TacticAI offers coaches valuable insights to enhance their corner kick strategies.
According to a research paper published in Nature Communications by Liverpool and Google DeepMind staff, the suggestions provided by TacticAI were preferred by human expert raters in 90% of cases over tactical setups observed in practice. These expert raters consisted of Data Scientists, a Video Analyst, and a Coaching Assistant from Liverpool.
TacticAI operates through a “geometric deep learning approach,” addressing key questions such as predicting outcomes for specific corner kick setups, evaluating the effectiveness of past tactics, and recommending adjustments to achieve desired outcomes. This AI system streamlines the analytical process, traditionally carried out by club analysts, by automatically computing numerical representations of players and facilitating efficient access to relevant past routines.
The researchers envision TacticAI being applicable beyond corner kicks, potentially providing tactical insights across various sports and industries. Petar Velickovic, a Google DeepMind employee and co-author of the paper, emphasizes the value of AI systems in identifying team errors swiftly, thus augmenting human capabilities rather than replacing them.
While AI holds promise in revolutionizing workflows within professional football, experts caution against the complete replacement of human input and insight. Patrick Lucey, Chief Scientist at Stats Perform, emphasizes the importance of effectively utilizing AI as a tool rather than a job usurper.
The collaboration between Liverpool and Google DeepMind has produced notable advancements in AI-driven analysis for football. This includes the development of TacticAI, as well as previous projects such as Game Plan and Graph Imputer, demonstrating the potential of AI to enhance various aspects of football strategy and analysis.
See also: Open-Source Initiatives Challenge OpenAI’s Dominance in AI Market
