Add Spokesman-Review on Google
By Dan ThompsonThe Spokesman-Review
A big baseball fan, Tim Plough saw the news last month that Major League Baseball had issued a memo to its teams, warning them, essentially, to use their dugout iPads properly.
According to reporting, published first by The Athletic, MLB teams may have been using the league-issued devices to access generative artificial intelligence platforms and to consult them about in-game pitch-calling decisions.
Plough, the UC Davis head football coach, said at the Big Sky football kickoff weekend in late July that he “wouldn’t be shocked” if artificial intelligence use creeped into football. Many coaches said that AI has already changed the way they prepare for matchups throughout the season.
But like other coaches asked about it, Plough pointed out that deciding what pitch to throw is a whole lot less complicated than calling a play for a team on third-and-4.
“When you’re talking about predicting pitches or predicting batters’ hot zones, it’s really one person communicating to one person, so that can be a huge advantage,” Plough said. “… (an at-bat) is such a one-on-one thing that if you have an advantage over one player that can change the entire course of a game.”
AI, as an evolution of analytics and as a timesaver in finding and organizing information, has already reshaped the way college football teams spend Sunday through Friday. Teams use it in recruiting. They use it to plan practices. They use it in film study.
But from the perspectives of Big Sky coaches and players, what happens on Saturdays in the games across the league remains, at least at this point, determined by forces that have been around since the start of the game.
“Whether it’s football, baseball, tennis, whatever it is, sometimes you like the pureness of the sport,” Montana head coach Bobby Kennedy said. “So hopefully we don’t get to the point where everything is about analytics and artificial intelligence and that there is still a human factor in this game.
“We’re always trying to move things forward and be innovative in our own way, but when is enough enough? That’s the question we all have to ask.”
A problem-solving tool, not a play-caller
Eastern Washington coach Marc Anderson has a reputation around the team as an early adopter and advocate for technology use. He’s been with the program since 2014 and is entering his second season as offensive coordinator.
“We use (AI) more internally to build tools that solve repetitive problems,” Anderson said last week.
He gave the example of calculating how many times the team ran a play in preseason camp. Previously, coaches had to do that by sorting through film or tallying it on a white board as they went.
“(Now) we upload our scripts and it tells us how many plays we’ve run so we can tell where we’re deficient and what we need to run more of,” Anderson said.
He also uses it to store variations of plays, which can be hard to remember over the course of the season. AI programs have also improved the Eagles’ opponent scout reports, Anderson said, which gives the team “a better starting point” in preparation.
“I think when you start to let it influence your decision making, I think that’s the human element,” Anderson said. “We still have to do our jobs. But I think letting it solve constant issues is not a bad thing at all.”
Eric Sanders, Eastern Washington’s defensive coordinator last season, left in the offseason to take the same job at Montana on Kennedy’s staff. In the few months he’s been coached by Sanders, Montana linebacker Peyton Wing has been impressed.
“Coach E (Sanders), he does enough film where I don’t think he needs to see AI,” Wing said. “He knows formations. He knows what the offense is going to do. He’s smart, man. I already trust him with that. He calls the plays, if he tells us to do something we’re doing it. And he’s got the look, and he prepares us right.”
Football’s complexity makes it both a prime candidate for the use of AI and also a massive challenge to the limitations of what AI can do. For years that’s been apparent in videogame adaptations when a human player competes against a “computer,” or AI player. One clear use of AI in videogames is the suggested plays it will give as someone decides which one to run.
But to Blake Weslin, a defensive lineman at Northern Colorado, there is something unsatisfying about simply accepting one of the suggestions.
“Here and there I’ll take one, but sometimes I think it’s better if I take it myself,” Weslin said. “It’s more fun that way.”
It’s one thing to suggest a play in a videogame. To Southern Utah head coach DeLane Fitzgerald, it would be quite another to do so in a real competition.
“I don’t see a path forward where AI creeps into playcalling,” Southern Utah head coach DeLane Fitzgerald said. “Man, I hope not.”
Fitzgerald, who has been a head coach at NAIA, Division III, Division II and now at Division I, listed off all the ways AI cannot match human capacities in football.
“Heres what AI can’t do. It doesn’t know the young man inside and out and what he’s good at and what he’s not good at,” Fitzgerald said. “(On) third-and 4, AI is going to give us a list: You should call one of these three plays. AI doesn’t know who got injured in the last series and is not in that position. AI doesn’t know the mental toughness or the physical toughness of the young man, or who’s tired, who’s not.”
‘I wouldn’t trust it’
Northern Colorado head coach Ed Lamb said he has hired an AI company to look at its possibilities. But he also is aware of the “battle of wits” presented if both teams are using AI in playcalling.
“I’m not aware of anything that would be able to predict both the opponent’s tendency in that moment, their desire to overcome the tendency, and recommend the correct thing to do,” Lamb said. “It feels a little bit like Sicilian logic when you’re trying to incorporate AI into that.”
And, when Lamb has looked into what AI offers, he’s not impressed.
“I don’t see that right now. I wouldn’t trust it,” Northern Colorado head coach Ed Lamb said. “We could easily pull up Chat and get some recommendations, but from someone who is moderately educated in football, it’s absurd, the suggestions AI spits out, at this point.”
Still, AI tempts coaches because it is in the nature of competition to be looking for an advantage, something technological advances have offered throughout the sport’s history.
When college football first allowed teams to use tablets in games on the sidelines to watch video of previous plays a couple years ago, Eastern Washington head coach Aaron Best warned his staff that he wasn’t going to have “an iPad conversation.”
“Let’s be organic,” Best said in July. “Let’s see the whole field and then look at the iPad to confirm what we saw and not just be a spectator and go to the iPad to be a coach. Coach in the moment and confirm what we see.”
Coaches can see plenty, too.
“Every defensive coordinator in the country on Saturday, he knows the offensive coordinator’s favorite six plays on third-and-short, third-and-medium, third-and-long,” Fitzgerald said. “Now, you’ve got to defend those six. But you can easily narrow it down to that.”
It’s the defending – the execution – that coaches and players pointed to again and again. Knowing what is coming, or could come, on any given play, is a problem addressed by preparation. But making the play is what settles the score.
At least for now, that human element is something coaches said AI hasn’t been able to match.
“In football, even though I know they might be running this defense and I know this play will work, I still have to get 11 people to do it right,” Plough said. “So even if I’m completely dialed in and know what’s happening, or the defense knows what play we are running, for them to get all 11 guys to understand that, to me that’s still the real art of coaching.”
