Football coaches are already using artificial intelligence to help them on the sidelines. Students at North Carolina State University built an AI that could predict whether an NFL team would pass or run the ball. The AI called the plays correctly 91.6 percent of the time during an NFL game. The technology may not yet be ready for a real-time game situation, but with more data and research into machine learning and game theory, AI could become a big league contender.

A play-calling AI would require big data about football. Big-data analysis company Splunk put 15 years of NFL data into its analysis platform to see if it could correctly predict next plays. During a live NFL game, the system correctly forecast the shotgun formation and direction of the pass. However, an even more advanced AI that relies on machine learning technology would need even more data, maybe even models like the physiology of each player.

Competitive football teams aren’t likely going to give that data away freely, though. Researchers at MIT have developed a technology that could be a way around that. The Eulerian Video Magnification process can measure a person’s heart rate or breathing pattern from a distance. This might be a way to identify which players are getting tired during a game. The technology can even be used to reproduce someone’s speech from afar.

A play-calling AI must also take the reverse psychology of the opposing team into account. Some teams will take a chance on a weaker play option if that’s more unexpected. That’s where more research into game theory comes in.

According to Wired, “game theory can guide teams on how to act in situations where they don’t know for sure how their opponents will behave.” If the coaches have a tool to help them predict the other team’s plays, they’ve practically won the game already.