Sports teams compile mountains of data–alongside human-generated scouting reports–in search of the next star athlete for their squads. A new generative AI tool called Scout Advisor, developed by IBM for Spain’s Sevilla soccer club, is indicative of how generative AI is becoming a key player in identifying future stars and may even help coaches on game day.
Sevilla FC, already well-known for its recruitment skills, is working with IBM to build Scout Advisor, an AI tool that leverages IBM’s watsonx’s natural language processing and foundation models to search, filter and analyze the massive amount of data present in the club’s databases. Sevilla FC’s data base includes quantitative information like height, weight, speed, number of goals and minutes played. There’s also a great deal of unstructured qualitative data from over 200,000 scouting reports. What Scout Advisor does is combine both to issue a summary of scouting reports alongside physical data to present a profile of an individual player based on stated requirements using natural language prompts.
There’s big money in all of this. Player scouting and recruitment at elite sports teams typically involve multi-million dollar investments and long-term contracts along with high levels of uncertainty and worries about return on investment, Past data analysis has been both time-consuming and restricted to a limited number of factors. Sevilla FC already has a reputation for early identification of rising star players which it has profitably sold on to other clubs. Scout Advisor increases the competitive advantage for Sevilla FC in this regard.
Scout Advisor “gives us a significant advantage in the recruitment process and enables us to find the best players for our team and improve our performance on the pitch,” says Jose Maria del Nido Carrasco, president of Sevilla FC. “We believe this collaboration will have a positive impact not only for Sevilla FC but for the sports industry as a whole. With this tool, Sevilla FC also demonstrates that technology is not just a goal but an intimate companion on the journey toward the future of our entity; it’s part of its DNA.”
Generative AI is likely to lead to an explosion of data scouting. Using generative AI, a club might ask for a list of players with certain metrics and quickly assess data and videos of the 10 best players at a certain position, for example, within the professional, collegiate or youth ranks.
Generative AI may still not be a catch-all for player evaluation, however. This is especially true at youth levels where other techniques like “bio-bending” that factor in the differences between calendar year and the biological stage of development for young athletes come into consideration. Youths generally perform better than their counterparts born between January and March. Another tricky area is “injury prediction” which requires access to confidential medical data and needs appropriate security guardrails.
Another key question is whether the player with all the star attributes will have his head in the game once it begins. Generative AI may be able to provide a clue for coaches. A company called Peak AI builds a personality profile of an athlete which can then be used as a baseline to indicate the athlete’s state of mind on game day. That pointer comes from a 30-second phone questionnaire that begins with a simple question: “Tell us about your day yesterday.” The verbal responses marry AI natural language processing with psycholinguistics. Peak AI says word choice and manner of speaking may indicate an athlete’s frame of mind at that moment. The AI is able to account for varying languages and accents as well as attempts to trick it.
Peak AI says coaches also can use its tools to evaluate an entire team’s state of mind at any given moment, describing itself as “an early warning system for the mind.” The company charges $99 per player per month. Peak AI says a number of European soccer clubs, the New Jersey Devils hockey team and Formula 1 race teams like Red Bull and Aston Martin have trialed or adopted its product. Peak AI anticipates a more widespread rollout in 2024.
