This article looks at the use of AI for developing athlete training plans
What’s the deal?
Artificial intelligence is here and is used every day by millions. We even use it for designing training plans, but can AI really handle the complex training needs of top athletes?
What did they do?
Researchers asked ChatGPT-4 to create adaptive weekly plans for elite swimmers and athletics (track) sprinters. Twenty-three coaches and 36 athletes reviewed and rated the plans on suggested frequency, intensity, and structure of the workouts using a 5-point scale. The workouts included different zones, ranging from easy recovery efforts to all-out sprints.
What happened?
Overall, coaches rated the plans as “neutral-agree”, whereas athletes were a bit more critical, rating them closer to “neutral-disagree”. AI did fairly well with easy, low-intensity sessions, but received lower ratings for more intense sessions. For swimmers, hard anaerobic sessions showed up too often. For sprinters, top-end speed work didn’t show up enough. About two-thirds of coaches said the plans were usable with small tweaks. But nearly half of the athletes wanted major changes, and one in four rejected the plans outright.
What can you tell me about the people?
The reviewers included experienced coaches and competitive athletes, both male and female. Overall, the coaches were largely in agreement with each other, but athletes’ opinions varied a lot more. Interestingly, older coaches and male athletes tended to rate the plans lower.
What else should I know?
A key component of any AI output is the strength of the initial prompts. The prompts used here were extensive and no-doubt built on leading industry knowledge of training processes, but improvements can always be made.
What will I do differently?
I have played around with AI-generated training plans in the past and have mostly been impressed, but there are always some eyebrow-raisers in there. I actually just purchased my first commercial marathon training plan, so for now at least, I’m sticking with the humans!
Acknowledgements
This research was published in September 2025 by L Puce and colleagues from the University of Genoa in Italy.
This article highlights our personal take on their research, but there is so much more information available! Check it out here and see what you think!
Abstract Athlete 