Football fans everywhere are gearing up to celebrate the sport’s most skilled athletes as they prepare for the start of another FIFA World Cup. But few get to see how the next generation of Messis and Ronaldos are discovered.
For decades, the beautiful game depended on the human eye: a scout on the sideline, attentively watching, waiting for that something special. That process, however, is becoming increasingly data-driven. Across elite football academies worldwide, technologies such as GPS trackers, automated video analysis, and AI-powered platforms are changing how players are identified and assessed.
In a sport shaped by money and global competition, the appeal is obvious. Clubs want to make better decisions and they want to make them as early as possible. The quicker a talented player is identified, the greater the potential return.
As my research shows, technology promises precisely this. Coaches and scouts are using new tools to make talent identification more efficient and, crucially, more objective. The question is whether talent can actually be reduced to numbers and whether technology and AI combat biases or create new ones.
Early identification
Modern football increasingly selects players at very young ages. Spanish footballer Lamine Yamal was scouted by Barcelona at just six years old; he made his first-team debut at 15.
This is not a necessarily a new development or consequence of AI and technology. David Beckham famously joined Manchester United as a 16-year-old.
Nonetheless, not all talent develops or is identified early. US star Alex Morgan only began playing seriously in her teens, Italy’s Luca Toni did not reach the top level until his early twenties; and England’s Ian Wright was nearly 22 when he signed his first professional contract.
So-called late bloomers are reminders that development is unpredictable. Especially in childhood and adolescence, athletes grow at different rates physically, psychologically and socially. This makes it difficult to compare players directly, and even harder to predict who will succeed.
AI systems, by contrast, rely on patterns. Neural networks analyse large datasets to identify what successful players have in common. The problem is that patterns are based on the past.
If previous systems have favoured certain types of players, for example, those who mature early physically or come from particular backgrounds (such as regional accessibility, higher socioeconomic status, having pre-existing profitable networks or even just parents who are involved and interested in sport), then AI may simply learn to reproduce those patterns. What looks like objectivity can, in reality, be a new form of bias.
Research has started to question whether AI can truly “solve” talent identification. While data-driven methods can support decision-making, they do not eliminate subjectivity. I have shown that decisions about what data to collect, how to analyse it, and what counts as “talent” are still shaped by human judgment and are, therefore, always subjective to some extent.
A second problem stems from the fact that data does not capture context. Just about everything can be measured, of course, but player qualities are typically revealed through interaction, experience and context – not metrics alone.
Studies have shown that coaches and staff struggle not with collecting data, but with interpreting it and turning it into meaningful insights. This clearly highlights that data is only useful if it is understood.
Fairness and diversity
Football has long struggled with unequal access and opportunity. Pathways into elite sport are shaped by social, economic and cultural factors. If AI systems are trained on historical data that reflects these inequalities, they risk reinforcing them.
Players who are physically more developed compared to their peers typically get an advantage. But this does not necessarily mean they have greater longterm potential. Wright or Toni, who reached higher levels at a later stage in life, would have likely been overlooked in such systems.
AI trained on such data may unintentionally prioritise early developers again and again. This would narrow the pool of talent rather than expanding it.
There is also the question of how data affects young players themselves. Previous research has shown that increasing monitoring can create intense pressure and a sense of constant evaluation and surveillance from above, for both athletes and coaching and scouting staff alike.
Football needs to decide what kind of talent it wants to recognise. Do we want to meet the new Ian Wrights or Luca Tonis? Or do we just want to copy and paste the same story, again and again?
AI has clear potential. It can help clubs process large amounts of information, identify patterns that humans might miss, and even widen access by allowing players to be seen from anywhere in the world. Used thoughtfully, it could support more informed and inclusive decision-making.
But we need to be clear about one thing: it is not a neutral tool and it is not increasing objectivity. It reflects the assumptions, values and biases embedded in the data and systems behind it. And it is chosen based on subjective ideas of what counts as valuable data.
The challenge, then, is not simply to adopt new technologies, but to question how they are used. Clubs need to be aware of the limitations of data. They need to invest in education and expertise. And they need to ensure that technology complements rather than replaces human judgment.
If the game becomes too focused on what can be measured, it risks overlooking what cannot. Without those safeguards, football risks losing what makes it special: its unpredictability, its diversity and that promise it has always proffered – that greatness can emerge from unexpected places.
- is a Project Researcher, Department of Sport Sciences, Malmö University
- This article first appeared in The Conversation
