Google Deepmind has just unveiled its newest AI model: TacticAI, an AI coaching assistant "that can provide experts with tactical insights" through "geometric deep learning" for Liverpool FC.

Announcing its multi-year partnership with Liverpool FC during its Natural Communications appearance on Monday, the tech giant promised "state-of-the-art results" on AI predictions for corner kick outcomes.

According to Google, 90% of the surveyed human experts favored its tactical set-ups for both practice and live games.

This includes capabilities to identify and analyze patterns in the rival teams and be able to respond quickly in real-time, like who is more likely to receive the ball or adjust tactics based on the players' current performance.

The AI was trained from 7,176 corner kicks in the English Premier League collected from 2020 to 2023 in hopes of "reducing the probability of a shot attempt for a defensive setup."

Also Read: Google's Gemini Produces Super Bowl Statistics Before It Takes Place

Google Deepmind Dips into AI Applications in Sports

Google has been hinting at intentions to integrate its Deepmind AI division into the sports industry as the hype for the technology continues to surge.

In the same collaboration announcement, Google said it first proposed the idea back in 2022 with its Graph Imputer model as a way to "downstream tasks in football analytics."

While the dedicated TacticAI may provide helpful suggestions to coaches in football, Deepmind's other released AI models might provide less reliable results, Premiere League or otherwise.

Google AI Models has a History in Unreliable Results

It can be remembered that its most notable AI model, Gemini, has earlier been reported as giving inaccurate analytics of the Super Bowl LVIII even before the game started.

This is on top of recent complaints of its image generator giving racially biased results of historical figures. Google has since paused the feature while it tries to fix the issue.

Google already advised users to not rely on the chatbot as it is vulnerable to "AI hallucinations."

Related Article: Google DeepMind Scientists Plan to Redirect Focus on Developing AI Model