When AI Gets Average: The Hidden Risk in Fitness Marketing
The promise and the problem with AI fitness marketing
AI fitness marketing feels like a breakthrough. You open a laptop and type a few lines. Within seconds, you’ve got a social caption, a blog post, or even an entire campaign. For a busy gym owner, that feels like time finally fighting on your side.
But there’s a growing risk that most don’t see.
Recent research from arXiv (2025) discovered an important insight. The study is titled “Forgetting and Regeneration in Large Language Models: The Regression-to-Mean Effect in Creative Advertising.” This research identified a key observation. It found that large language models do not actually create. These models, such as ChatGPT, Claude, and Gemini, replicate existing patterns. They replicate the most common human patterns. Over time, that makes your marketing sound polished but predictable.
It’s the perfect example of regression to the mean—and in fitness marketing, average doesn’t grab attention.
Why regression to the mean hurts creative gym marketing
“Regression to the mean” sounds academic. However, its meaning is simple. When systems repeat the same process often enough, their results drift toward the average.
That’s exactly what happens in AI fitness marketing.
Each time an AI model writes copy, it predicts the next word based on probability, not imagination. It picks the safest, most common phrase in its data set. That’s why gym ads generated by AI all start to sound alike—full of transformations, journeys, and limitless possibilities.
Those words work fine in isolation, but collectively they flatten creativity. And when your brand voice flattens, your message fades.
Inside the research
The study tested creativity using over a thousand real ad ideas. Researchers stripped each ad back until only basic product facts remained, then asked AI to rebuild the original creative version.
The results revealed how quickly AI fitness marketing loses its spark:
- Emotion disappeared first, replaced by neutral statements.
- Visual detail vanished; the messy humanity of sweat, noise, and light was erased.
- Longer text didn’t mean better ideas. AI simply padded safe words around the same structure.
The models produced work that was fluent but forgettable. The more they “expanded,” the closer they got to mediocrity.
Why average is the enemy in AI fitness marketing
Most gym owners think average is acceptable—if posts are clean, readable, and grammatically correct, that’s a win. But in marketing, competence is not connection.
The fitness industry runs on emotion. People don’t join because of logic; they join because something feels possible. AI doesn’t feel. It can mimic emotion, but it can’t create it.
Behaviourally, this is tied to the salience bias—the tendency for people to notice what stands out. When every gym relies on the same AI phrases, nothing stands out. “Join today” scrolls past like wallpaper. The research on cognitive overload in marketing makes a related point: when everything looks and sounds the same, brains stop processing it entirely.
Average is not neutral; it’s invisible.
The confidence trap in gym marketing AI
One reason artificial intelligence in fitness feels so convincing is its confidence. LLMs write cleanly, with perfect punctuation and rhythm, so it reads as professional. But that fluency hides a lack of understanding.
In psychology, this mirrors cognitive ease—our bias toward what feels familiar. Smooth, predictable language feels right even when it’s empty. That’s why AI copy gets quick approval from human editors but low engagement from audiences. Research on processing fluency shows that ease of processing builds trust, but when every competitor has the same fluency, the advantage disappears.
The study measured this “averaging” effect using linguistic diversity scores. As AI generated more copy, vocabulary shrank. Every ad became a variation of the same safe pattern.
What this means for your brand voice
Your members don’t need another generic promise of transformation. They need your voice—your local humour, your story, your quirks. That’s what cuts through in AI fitness marketing: authenticity mixed with behavioural insight.
When you hand the microphone to a model that’s trained on global averages, you lose that local soul.
Imagine every gym in Victoria using the same AI tool. Within months, everyone’s Facebook posts start sounding eerily alike. The slogans differ, but the rhythm is identical. That’s regression to the mean in action.
How to keep AI fitness marketing creative
This research doesn’t say abandon AI—it says guide it with intent.
The models perform better when given creative “markers”: small cues that tell them what kind of human detail to include.
Here’s how to make AI fitness marketing distinct:
- Feed it a story, not a slogan.
Tell it about your members or your space. Describe a single moment—like the first splash at 6 am. It could also be the clang of a barbell in winter. - Use emotion markers.
Ask for a feeling: relief, pride, quiet determination. It helps the AI ground language in experience, not clichés. - Ban the buzzwords.
Create a “not allowed” list—transform, limitless, crafted, journey. Make the model work harder. - Human edit for truth.
Use AI to start, but not to finish. The best ideas still come from lived reality—your staff, your stories, your floor.
By using markers, you push the AI outside the average lane and back toward creativity.
The psychology behind the problem
The regression-to-the-mean effect in AI advertising mirrors our own behavioural shortcuts. We naturally drift toward what feels easy and familiar. That’s safe—but safe rarely sticks.
The human brain remembers difference, not repetition. The Von Restorff effect proves that distinctiveness increases recall. In marketing, being slightly unusual is what keeps people talking. The same principle applies to what makes a fitness slogan stick: the memorable ones break patterns rather than following them.
AI fitness marketing can support that goal—but only if humans set the boundaries. You still need to inject the emotion, the humour, and the friction that make people stop scrolling.
The hidden risk of average
The biggest risk with AI fitness marketing isn’t that it writes something bad. It’s that it writes something good enough.
“Good enough” lulls you into comfort. You publish it. Engagement looks fine. The numbers seem stable. But slowly, your distinctiveness fades, and your brand becomes another background voice in an overcrowded feed.
AI makes it easier than ever to sound professional—and easier than ever to sound like everyone else.
Key takeaway
AI fitness marketing can save time, but it can’t save your voice.
Use it as a smart assistant, not a storyteller. Guide it with emotion, imagery, and behavioural science. Because in the end, the safest message is also the one people forget first.
Average gets ignored. Humanity gets remembered.
Bibliography
- Chen, Y., et al. (2025). Forgetting and Regeneration in Large Language Models: The Regression-to-Mean Effect in Creative Advertising. arXiv preprint arXiv:2509.25767.
- Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux.
- Loewenstein, G. (1994). The Psychology of Curiosity: A Review and Reinterpretation. Psychological Bulletin, 116(1), 75–98.
- Berger, J., & Milkman, K. L. (2012). What Makes Online Content Viral? Journal of Marketing Research, 49(2), 192–205.
Discover more from Fitness is BS.
Subscribe to get the latest posts sent to your email.
