The Language Barrier Behind FIFA's AI: Teaching a Machine What a Cross Is
The collaboration reveals that integrating AI into sports depends less on raw computing power and more on translating tacit domain knowledge into formal definitions that a machine can act on.
Reporting from 1 source: ASCII.jp.
A new article from ASCII.jp details the initial challenge of FIFA and Lenovo's collaboration: defining basic soccer terms like "cross" in precise, machine-readable language so the AI tool could understand the game. The process required workshops to align soccer analysts with AI engineers, and Lenovo assigned "bridging" staff to translate between business and technical sides. The result is the FIFA AI Pro tool used by all 2026 World Cup teams.
FIFA and Lenovo's partnership to build the FIFA AI Pro tool began with a problem that had nothing to do with code. The AI had to learn what a cross is-not the way a human does, by watching a game, but as a set of precise conditions about player position, ball trajectory, and intent. The same went for pass, shot, and press. The article notes that player body data is captured at 50 samples per second (29 data points per player) and ball data at 500 samples per second, but raw data alone does not make a tactical AI. FIFA analysts and Lenovo engineers ran repeated workshops to tie those numbers to concepts like formation, pressing trigger, and transition pattern. On the human side, Lenovo assigned a bridge role: staff who could explain what the technology could do, not how it worked, to FIFA's non-technical business leads. The tool now serves every team in the 2026 tournament.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.