Artificial intelligence (AI) is the hottest — and most unsettling — topic shaping the future of business, and college sports is no exception. Decision makers, staffers and coaches know it’s changing the world, even if no one fully understands how. Like any industry, college athletics is caught between excitement and unease: Will AI become a recruiting tool, a scouting assistant, a business optimizer—or something else entirely? For now, it can already do things as simple as creating the very image attached to this story.
To find out how AI is actually being used on and off the field, CBS Sports spoke with more than a dozen people inside and outside athletic departments about what’s real today — and where it could go tomorrow.
IBM defines AI as “technology that enables computers and machines to simulate human learning, comprehension, problem solving, decision making, creativity and autonomy.” The latter is the work we’ve all seen in science fiction movies like I Robot or current trendy news stories about people falling in love with their chatbots. But part of the confusion surrounding AI is that its current marketing falls quickly into the buzzword territory. To those on the data science side, there are things under the AI umbrella that have a specific use case like a “large language model” (LLM) that is the foundation for a product like ChatGPT, or computer vision which allows machines to view moving objects and translate them into a digitally rendered image like CBS’ Romovision during NFL broadcasts.
Then there is a more nebulous term like machine-learning, which is used to market products to consumers using AI.
But where is AI’s use as it stands right now?
Look no further to the front office sector of college football teams. As front offices expand, schools are bringing in people outside of traditional backgrounds to fill roles. Whereas a coach most likely played the sport at some level before getting into coaching, it’s not at all uncommon to find a GM or front office staffer who didn’t play at all. Sometimes those people can have an analytics background or background from the recruiting operations world.
AI isn’t at a point where it’s recruiting for your favorite team with a chatbot posing as a coach, but there are small ways people use AI. One Big Ten recruiting assistant said he uses ChatGPT to check the text messages and graphics they sent to players; another staffer says he uses it to synthesize a scouting report into a quick bulleted list to give to the coaches he works with.
Benjamin Elsner, formerly chief of staff of website The 33rd team, thought he was heading to Gainesville, Florida for a professional development trip. He ended up interviewing for and being hired with the title of director of football strategy as the Gators expanded their front office staff. His stated job description includes “leading the program’s efforts in analytics and AI for all parts of the organization.”
“I do think ChatGPT can be really helpful in a couple of different ways,” Elsner told CBS Sports. “One of the ones that pertains to me from a data science perspective is it can be very helpful in writing code. Now, it’s not going to write the code for you, but if you’re running some bugs or things along those lines, it can be very beneficial or creating more efficient code. It’s very, very good at being able to help with commands on Excel as well. So I think there’s a lot more ways that we can continue to use it in the future, but in the first six months here, I’m pretty proud of the extent to which we’re starting to use it.”
Nebraska’s Year 3 breakthrough under Matt Rhule still out of reach as Huskers can’t shake old script
Cody Nagel
Nebraska is taking things to another level with the use of AI agents, spearheaded by Sean Padden, the Huskers’ assistant AD for strategic intelligence. An AI agent is an autonomous workflow that can work around the clock to, for instance, scour social media for news about a certain player in order to help a program make a more informed decision and move quickly during the all-important transfer portal windows when negotiations are compressed sometimes to a matter of hours. Padden describes the challenge of portal scouting as narrowing 14,000 names to four in order to figure out who to pursue in order. The specific way an AI agent can help is with filling in the gaps to simple questions with not-so-simple answers during the portal process, such as: why did a player miss a certain amount of game time over the course of a given period? Injury? Poor play? Off-field trouble?
“Right now, we are in version 1.0 and so right now I am just sort of auditing it by putting in guys that I knew missed time, and seeing what it’s coming back and telling me, and seeing if it’s catching the time loss,” Padden said. “There’s agents that are looking at high school history, recruiting history, generic news articles. The goal is to be able to find everything that we can. You have to hope that the AI catches the time loss and is curious about it, and hopefully it can give you an answer as to why there was missed time there.”
Padden likens the process of fine-tuning the agents to a sailboat. He points the agents in different directions and tries to see where it goes. It’s not ready for prime time yet, but he’s hoping one day it’ll be at a place where his coaches can have it on their phone to deliver accurate answers about players they’re targeting at a moment’s notice. Schools at a level like Nebraska have different needs than a Group of Five program, however, which is where the duality of AI can be used to fill needs:
- A program the size of the Huskers theoretically has their pick of just about any athlete anywhere, and they’re trying to narrow the pool of athletes down to a reasonable number.
- A smaller school is trying to expand their pool as wide as possible to catch players who have slipped through the cracks. It’s possible for both levels of programs to use AI for their needs.
There are other ways inside athletic departments that AI can be deployed.
“What we have found is that AI really helps you find some hidden incremental dollars in existing [revenue] streams,” Arizona athletic director Reed-Francois told the JohnWallStreet podcast in a recent interview. “Just scratching the surface, don’t pretend to know all the answers, but I’m fascinated by it and I love the tool and I feel like there’s so much more than we can even do.”
Reed-Francois says that Arizona isn’t quite at the point of using it on the court or the field but the school has hired a campus AI officer, David Ebert, to help figure out how to use the technology on the business side.
As schools figure out what they’re doing with AI, various vendors in the football space are as well. In 2024, Hudl, best known for its near monopolization of high school game film, acquired the company Statsbomb. In May, the company announced a new product: Hudl IQ, a data and analytics platform that the company claims uses AI to tag formations, routes, coverage, and blitzes in game film to help recruiting, game prep and self scouting. Not to be outdone, Teamworks, a software company primarily used for roster management, bought Telemetry (a leader in the player tracking space using computer vision) and Zelus (a company that delivers multiple types of predictive models to aid roster building). Teamworks recently announced a round of fundraising to “accelerate AI-powered innovation,” highlighting the ability to attract investors eager to tap into the apparently limitless potential of AI.
Teamworks and Hudl are eager to sell their services across college and professional sports, and it does not come cheap. Teamworks recently became a presenting sponsor for the annual personnel and recruiting symposium where over 1,000 people from 156 schools and 24 NFL teams come together for networking and professional development. Teamworks delivered an hour-long seminar to attendees about its services. Costs for Teamworks’ basic package for a football team is in the mid-five figures per year but can stretch into six figures with various additions or uses across multiple sports. As college sports become increasingly professionalized, personnel departments have to factor in that every dollar spent could go either to an athlete via rev share payment or to something else.
There’s also the practical day-to-day calculus anyone has to make when using a piece of software. Is that piece of software actually useful, or is it just software for software’s sake? Is it an AI problem or is it a database problem? Let’s say you want to filter every wide receiver in America who runs a 4.6 40-yard dash or faster. If you have the wideouts, and you have the 40 times, there are myriad ways just in excel for instance, to get those answers. Where AI can come into the picture is going layers deeper to provide answers to deeply contextualized questions.
For instance, you can you ask a personalized in-house AI tool: Of the wideouts in the SEC who have at least two years of eligibility remaining and have contract demands in “X” range, which of them do we have scouting reports on, and which of them have missed fewer than “X” number of games due to injury. Fiddling with menus may not be something intuitive for a coach, particularly an older one. But could an older coach use speak to text to ask an AI tool that and get the answer within seconds? That’s some of the desired end result for new-school data scientists within programs.
It’s important for vendors to have people in their organizations who actually know what coaches and front office people want. That’s why they’re hiring them. FLX Sports, the AI-powered sports tech company that aims to connect brands to athletes in the NIL era, just hired James Thomson, who was formerly the director of internal scouting at USF. Thomson joined FLX Sports as executive director of football to help bridge the gap.
“It’s very hard for tech people to speak to understand the sports world,” Thomson said. “Obviously, sports people aren’t built to, you know, engineer software, so you really need the mixture of both to have a good company. And that’s what separates FLX. I’m able to kind of guide them with what the climate is really like, because there’s perception, there’s media, there’s there’s vantage points from parents, there’s vantage points from coaches or from fans, but to really understand the inside of a program is kind of where I add value.”
But there are companies assisting college football teams that aren’t directly selling an AI tool as part of their services, but rather using it on their own backend to power the services they provide to schools.
Players Health, a sports insurance provider that works with nearly 30 schools across the power conferences, uses AI agents to scrape public and private sources to model an athlete’s risk of injury and transfer. They offer insurance policies for schools to protect them against loss of value when an athlete gets hurt and can’t fulfill their revenue sharing contracts.Their list of sources include Pro Football Focus, Sportradar, ESPN, and Players Health’s own internal data.
Teamworks and Players Health recently partnered to integrate insurance quote generation capabilities into Teamworks software for college programs. Athletic departments receive an actuary table that includes injury risk and transfer risk by position group and a writeup by position group that provides context for each group, calls out snap volume, red flags and recruiting areas of importance. With a school they work with they might identify that multiple running backs have an injury score that puts them at risk, so perhaps they might want to prioritize a back in the portal. In the future, perhaps there will be a more robust visualization generated.
“When you go play EA [College Fooball 26] and you’ve got the injury factor on a player that’s going off, like, okay that’s all cool to play in a video game, but that’s actually a very real aspect of how roster management and a season management should be looked at, and that is part of what we kind of have as our North Star is always looking at,” said Tate Gillespie, the Players Health VP of Collegiate/NIL Strategy. “I know it’s in a video game, but that’s pretty dang real, right? Like, that’s how you should be approaching each player. You should be looking at your roster and being able to project that.”
But is there a future for AI on the field?
There are white papers on the subject like QB-GPT and others promising that in the not-too-distant future, functioning models will exist where a coach can plug his playbook, scouting report, and the opponent’s tendencies and be able to have a functioning model beside them to gameplan in the palm of their hand. There are also enterprising coaches who embrace technology that are toying around with how to automate gameplanning. Some, however, are certainly not.
“I’m not using ChatGPT on my sideline to decide what we’re gonna run,” Oregon head coach Dan Lanning told reporters before his team’s season opener in August. “I’d be interested in what kind of information that provides, but I don’t know if that’s made it to us. … AI is something that’s going to continue to grow. It’ll continue to be a part of the game, probably as much as anything, from a scouting standpoint. But, it’s probably still the early days for us at our level.”
The practice of claiming a certain technological innovation is 5-10 years away has become cliche in Silicon Valley. It hits a sweet spot of being just far away on the time horizon to seem distant, while also making sure few, if any, people remember the predictions one way or another when the time comes. Many AI initiatives across industries fall into that bucket due the goldrush mentality Wall Street investors seem to have around it. But as is common with the promises software companies espouse, are they just repackaging technology that already exists?
“I think the program is (the video game) Madden, and you could probably do it by the end of the day today,” Padden said. “They could generate an entire freaking game for you. You want to see the first 15? They just did it. So I think that as far as play calling, as far as coaches go, there have been analytics for as long as there’s been cut ups, maybe even longer. And so I don’t think that football needs to catch up in any regard. I think that we’ve always been interested in the way the game gets carved up by our opponents. And hey, I may just speed some things up. But you could auto play a whole game on that, and then there you go.”
AI won’t replace the players, and it’s dubious to what degree it can even be used to assist coaches — though there are plenty of fans who today would take a computer over their beleaguered offensive coordinator — but in one way or another the technology will find its way into college football’s on-field product, just like like sideline tablets and helmet radios have. It’s only a matter of time.
Read the full article here