Advertisement
Supported by
A.I. fitness trainers are here, at a fraction of the price of human coaching. But how useful are they?
Send any friend a story
As a subscriber, you have 10 gift articles to give each month. Anyone can read what you share.
Hiring a personal trainer is one of the best ways to stay consistent with your workout, push your limits and try new things. But it often costs more than $100 a session, and getting to and from the gym can be time consuming.
Over the past decade, fitness apps have been attempting to replicate the personal training experience, and in recent years many have incorporated artificial intelligence to generate workouts. A widening array of products offer custom workouts based on your abilities, goals and available equipment, for much less than a personal trainer (typically around $100 a year).
However, some experts warn that, while A.I. fitness apps are useful for many exercisers, they’re not appropriate for everyone.
A.I. fitness coaching apps create personalized training programs based on your goals, the type of workouts you want and your equipment. Unlike ChatGPT, which uses enormous amounts of existing data to predict the next word in a sentence, these apps take an individual user’s data and employ algorithms to create personalized workouts, based on other users’ experiences.
As you complete workouts, the app customizes your next training sessions, incorporating feedback you provide about how you felt and your performance — how long it took, how much work you did or how much weight you lifted.
Many of the apps also integrate data from fitness wearables or smart watches. In the last couple of years, fitness apps have been incorporating more complicated blends of heart rate, mileage and calories burned to create new workouts.
One of the oldest and most popular of these, with more than 55 million subscribers, is Freeletics, which incorporated a form of A.I. into its program in 2017. One benefit of such a large subscriber base is that it has an extensive data pool from which to create workouts, helping the A.I. predict what training would work best for you based on how others performed, said Confidence Udegbue, an executive who directs the development of new products at Freeletics. The app costs about $100 per year.
A host of other apps have recently launched or incorporated A.I. into their programs and now offer similar services. Fitness AI, Aaptiv (which is launching a new coach product in May) and FitBod all cost slightly less than $100 a year. JuggernautAI, which has a strength-training focus, costs $350. Some smart home gyms, like Tonal and Tempo, include proprietary equipment with motion sensors that use A.I. and cameras designed to track your movements and offer feedback on form. Those cost thousands of dollars, plus subscription costs.
In the future, training apps could collect even more data to measure effort — in addition to heart rate and respiration, perhaps eye dilation or oxygenation — and integrate it with what we know about exercise science to give ever more personalized feedback, said futurist and astrophysicist David Brin. But, he added, more data doesn’t always lead to better advice.
“What’s at issue — as always,” he said, “is whether the output advice will actually be good for you, over the long run.”
If you’re familiar with the movements an app might suggest, and you’re self-motivated, A.I. trainers are a cost-effective way to get a customized training plan and shake up your workout routine. A part of personal training, especially strength training, is thinking through formulas about how many times you perform an exercise (the reps), how many groups of those reps you complete (sets) and how those numbers should change as you progress.
“What A.I. does exceptionally well is determine sets and reps,” said Cooper Mitchell, the owner and founder of Garage Gym Reviews, who has spent the last 10 years reviewing home gym equipment and technology, including A.I.-powered fitness apps.
But if you’ve never squatted or bench pressed before, he said, it’s best to learn these movements under the watchful eye of an experienced personal trainer.
Furthermore, computers can learn a lot about different fitness regimens, but they can’t yet replicate the social interactions that make training effective, said Nikola Banovic, an assistant professor of computer science and engineering at the Artificial Intelligence Laboratory at the University of Michigan who specializes in human-computer interaction.
“Just because the A.I. is saying you need to run more,” he said, doesn’t mean you’re “actually motivated to do it.”
This is especially true for veteran athletes, who often say the toughest battles in fitness are psychological, not physical.
“The coach becomes a little bit of therapist. I text my coach just as often to complain to him as I do to give him the results from training,” said David Tao, co-founder of BarBend, a website about strength training and strength sports.
The existing A.I. apps can’t provide encouragement or comfort you on an off day. However, Dr. Banovic said, in the near future they may use programs like ChatGPT’s to talk about your workout or offer motivation. Even then, it will be important to consider the value of personal interaction, Mr. Mitchell said.
“As humans, we need more than just the workouts that will get us to our goals,” he said. “We need to be inspired and encouraged — that’s something only a real trainer can do right now. I think it will be some time before A.I. training does this well.”
Hilary Achauer is a freelance writer focused on fitness, health, wellness and parenting.
Advertisement
