My cousin is a club-level cyclist. Nothing professional — he rides for the joy of it, and competes in regional events on weekends. Last year he started using an AI coaching app, and his performance has improved more in six months than in the previous two years.
When I asked him what changed, he didn't say anything about the model or the data. He said: "I stopped feeling like I was making it up alone."
That's the story of AI in sport that interests me. Not the pro athletes with data teams the size of a small company. The weekend warriors, the high school teams, the local coaches — people who now have access to insights that used to be reserved for elite environments.
Where AI is already on the pitch
Video analysis. Every Premier League team now has AI systems that chop up match footage by player, by event type, by tactical situation. Coaches can ask for "every time we lost possession in the attacking third" and get a clip reel in seconds. This used to be three people with a remote and a clipboard.
Performance tracking. Wearables plus AI give coaches continuous visibility into athletes' training load, recovery status, and injury risk. The good implementations catch overtraining before it becomes an injury. The best implementations respect the athletes' agency — showing them the data, letting them have a say in what it means.
Tactical analysis. AI systems looking at positional data can quantify things that used to be intuition — how compact a defence is, how well a team presses, how much space a particular player creates. Coaches argue about the interpretations. That's a healthy argument.
Officiating. VAR in football, Hawk-Eye in tennis and cricket, automated strike zones in baseball. The technology has its critics, but the trend is clear: officiating decisions that used to be subjective are becoming objective where they can be. The sports still have their controversies, just different ones.
Talent identification. This is both exciting and concerning. AI systems analysing youth-level footage can spot future stars earlier. Done well, this helps kids from under-resourced backgrounds get noticed. Done badly, it risks narrowing what "looks like an athlete" in ways that might exclude late bloomers.
Where it's reaching amateur sport
Form correction via smartphone. Apps that watch your running form, your tennis swing, your yoga poses using your phone's camera, and give you real-time cues. The first generation was gimmicky. The current generation is genuinely useful.
Adaptive training plans. My cousin's cycling app looks at his recent rides, his recovery data, the weather forecast, and his target event, and adjusts the next week's training plan. It's no substitute for a great human coach, but for someone who'd otherwise have no coach at all, it's transformative.
Community and competition. AI-powered matchmaking in online games and competitive apps lets players find opponents at their level. Not flashy. Keeps more people in the game, which keeps them coming back.
Injury prevention at the club level. Wearables and simple AI models can flag a young player whose training load is suddenly spiking in a pattern associated with stress fractures. This kind of intervention used to require an elite medical staff. Now a volunteer coach at a Saturday-morning club can get the same insight.
The shadows
Surveillance creep. The same tools that help professional teams monitor their players are being sold to amateur leagues, corporate wellness programmes, even secondary schools. The privacy implications get very serious very quickly when we're talking about children's bodies and health.
Commodification of talent. When every teenage athlete can be scored by AI, the pressure to optimise them starts early. Kids who'd otherwise play multiple sports casually get funnelled into single-sport specialisation younger and younger. The research is pretty clear that this is bad for their long-term development, and for their enjoyment of sport.
The quantified-self trap. Not everything about sport should be quantified. The part where you just enjoy the sun on your back, or the camaraderie of your teammates, or the weird satisfaction of a Tuesday-evening session in the rain — that's not a number. AI can unintentionally push people towards treating sport as performance, when it's also (maybe primarily) meant to be play.
What I'd like to see more of
- AI that respects the athlete's voice. The best systems treat the athlete as a partner, not a subject.
- Tools designed for small clubs. Most AI sport tech is built for Premier League budgets. Most sport isn't that.
- Privacy-first design. Especially for anything involving young people. Local processing, minimal data, clear rights.
- Appropriate limits on surveillance. The locker room, the showers, the recovery days — not every corner of an athlete's life needs a sensor on it.
A small, unglamorous celebration
Amateur sport is where most of us meet the joy of movement. It's where friendships are made, kids learn how to lose gracefully, and adults remember how to play. AI that makes this accessible to more people, or that gives people better coaching than they could otherwise afford, is doing something quietly good.
My cousin is still riding for the joy of it. The AI didn't change that. It just gave him a more thoughtful training partner, when his human ones were busy.
If you're thinking about AI in a sport or fitness context, we'd be happy to help you think it through. Especially the accessibility angle. That's a favourite.
