This blog is mostly for my fellow EMS trainers. At this point, I think we’re all well aware of the pros and cons of current AI models. I’ll skip the benefits here; they are obvious.

The cons, however, are important to keep in mind. We know that AI sometimes makes mistakes, goes off the rails, and simply makes things up. In this post, I want to highlight what you need to watch out for when creating AI content about EMS—whether for marketing, blogs, IG posts, or website copy.

The Data Problem AI needs to be fed large amounts of data to compile accurate information. Unfortunately, there isn’t actually that much high-quality information about Whole Body EMS (WB-EMS) available publicly.

Here is where AI pulls its data from, and why that’s a problem:

• Manufacturer Websites: Manufacturers aren't necessarily trying to educate the world; they are selling to interested parties. Consequently, their content is often skewed toward marketing bullet points rather than scientific nuance.
• Research Studies: This is the trickiest area because AI struggles to differentiate between the three main types of “EMS”:
  
  • Localized EMS: This has been around for 50+ years and focuses on one or two joints. There are many studies on TENS units, EMS for rehabilitation, muscle recovery, and some where localized EMS was used for muscle development.
  • EMSculpt: A passive treatment often backed by manufacturer-funded studies. Be skeptical of the research here— Any time you see an equivalence of minutes under stimulation to many sit-ups, you can dismiss it as “false advertising”. That is not how EMS works with the muscle.
  • WB-EMS: There are about 150+ valid research studies on WB-EMS specifically. I have a Google alert set up that sends me links to any study that mentions “WB-EMS” and I have been reading them for years.
• Blogs: Some of us write blogs about EMS—I’m one of them (clearly). I use AI to help write some of them (mostly to explain general physiology like sarcopenia or diabetes), but I never use it to explain what EMS does or how it works because of the issues I’m mentioning in this blog.

The Infinite Loop of Doom. Here is the core issue: Manufacturer websites are usually created by marketing teams using AI to write copy. Blogs are often written by AI as well.

This creates an "infinite loop of doom" where people use AI to create content, and AI then scrapes that content to learn, assuming it is fact. If an AI makes a mistake, it gets perpetuated and further skewed. This isn't specific to EMS, but because our field is small, the errors are more obvious.

The Hallucination Problem AI often fails to distinguish between Localized EMS, WB-EMS, and EMSculpt. Worse, if you ask for research and it can’t find any, it will sometimes hallucinate (make up) studies entirely.

As I mentioned, I keep a close pulse on WB-EMS research. Recently, I wanted to write about EMS and menopause. I hadn't seen any studies on this, but just to be safe, I asked ChatGPT to find some.

To my surprise, it pulled up five studies, complete with summaries and outcomes. When I asked for direct links/citations to double-check them, it provided five links. None of them were real. They were random links to unrelated studies, and the outcomes were completely fabricated.

The Contradiction Issue Another frequent AI blunder is stating a "fact" in one sentence and then stating the complete opposite in the very next sentence.

A prime example is the distinction between FDA Approval and FDA Clearance. AI will frequently conflate the two terms or use them interchangeably within the same paragraph, despite them having very different legal and safety meanings. You can check my previous blog specifically about FDA approval vs. clearance to see exactly what I mean.

The "90% of Muscle" Myth Another common AI blunder I see on many posts about the benefits of EMS: "EMS activates 90% of muscle."

It sounds scientific, but it’s wrong. The correct phrasing is that EMS activates 90% of muscle fibers within a given muscle group—and WB-EMS only does so under certain frequencies and pulse widths. That is a massive difference in meaning.

Recommendations for Trainers If you are using AI to help with your marketing, follow these rules:

1. Be Specific: Use the prompt "WB-EMS," not just "EMS."
2. Verify Citations: Always ask for links to research studies, then click them to ensure they are real and actually about the topic at hand.
3. Fact Check: Ensure the logic aligns with what you know as a professional. AI often contradicts itself from one sentence to the next
4. Avoid Specific Numbers: Personal preference, but stay away from claims like "equivalent to 4 hours in the gym", “engages 90% of muscles”, or "20,000 sit-ups." In many cases, these specific metrics are bound to be inaccurate.

This blog was written by a human, me!

If you're interested in EMS training with me and live near Santa Monica, Brentwood, Palms, or Bel-Air, please send me an email at conradfitness@gmail.com or book a 15 min call with me.

I also offer in-studio sessions in Beverly Hills, located at 300 S. Beverly Dr., alongside my colleague Sarah, who provides private and semi-private Pilates sessions. Find out more here: lamaisonpilatesems.com

If you are looking for an EMS trainer in other cities, please check my trusted network.

‍