Only 10% of people prefer AI coaching: here’s what it gets right and wrong about the future of fitness.
I saw a recent industry survey that said only about 10% of people would choose AI workout guidance over a human coach. Most people still lean towards a real person, especially when it comes to motivation, connection and the experience of being coached. A lot of people read that as a rejection of AI. I don’t see it that way. I think it mostly comes down to what people mean when they say coaching, and what they’re actually expecting AI to replace.
Someone who writes you a programme and gives you an “atta boy” every Wednesday and Saturday morning. Someone who just gives you a series of exercises. That’s not a coach. That’s a technician. And why would you do anything other than have AI for that?
If I’m a member of a mid-market gym and I know what I’m doing, I’d much rather have AI in my pocket to give me a workout than a person. I don’t need a person for that.
Why lived experience still matters
But if I want coaching and community, I want somebody to take me through it who’s got lived experience.
You’ll never be able to teach AI what it’s like to go through that workout. I’ve been through every workout system there is. When was the last time you saw AI do a drop set, increase the eccentric on their RDLs for eight reps, and understand what that does to you? You have to have been through it.
I’m never going to trust a robot to give me its interpretation of what it’s like to do RDLs at that speed. There’ll never be that sense of understanding or appreciation.
AI is brilliant at giving people information, organising information, which is now commoditised, and putting structure around workouts in micro and macro cycles. The same with nutrition coaching in terms of macros and food plans. All of that has already gone to AI. But that’s not coaching.
Where AI will take people
I think AI will be adopted by more seasoned gym-goers, people who already know what they’re doing. I think we’ll lose people to AI over time. They might start in a gym, learn what to do, then move to an AI-driven programme.
But AI will still need to be underpinned by a brand or a person. If it’s a brand, it’ll be at scale. As soon as AI health tools land with credibility at around £19.99 a month, you’ll have all the information and guidance you could ever need in your pocket.
At that point, you’re really paying for implementation and experience.
On your own, even following an AI programme, it’s a bit like tickling yourself. It’s just not the same.
I’ve trained on my own, done my own thing, had a squat rack at home. Over time, it just gets progressively less good.
Gyms are about energy, sweat and experience. Wellness is a mental state. Gyms are physical places. Sitting in your front room following an AI-driven programme isn’t the same.
AI is going to tip the industry somehow, but I don’t know where we’ll be in 20 years. I don’t think anyone does. You can make a prediction and look smart if you get it right, but chances are you won’t. I also think that GLP-1 drugs will probably have more impact over the next one to three years than AI.
What changes for coaches
Programming and raw knowledge become less important.
The coach of the future will be a better communicator, with better self-insight and better empathy. They’ll be better human beings, focused on immersive connection.
One of the upsides of AI is that it weeds out people who only know how to programme. AI already supports us. It accelerates what we do.
From an operator’s point of view, the benefit is that it lets us concentrate on what we actually do well, which is training people.
With nutrition coaching, we’re not even sure it’s worth doing ourselves. We might just let people use AI. Do we train an internal AI with our philosophy, because people trust us to curate information? Or do we just let them go straight to AI health tools?
Someone can already say: I’m 100 kilos, I want to be 90 kilos in 16 weeks, this is my training schedule, these are the foods I like, this is where I shop. Can you do my shopping list, create my menu, and give me the recipes? AI can already do that.
As operators, we need to dial down on the bits that only we can do.