Training | POSTED March 26, 2023

I Asked ChatGPT to Write Me a Training Program and This Was the Result

Like us, you’ve probably read a lot in the news about how powerful AI (Artificial Intelligence) is becoming, and all the cool things that it can do. Simply put, it is the simulation of human intelligence in machines that are programmed to think like humans and mimic their actions. Heck, you can even code with it! We recently started playing around with a new AI system called ChatGPT, and asked it to write us a training program. The results were pretty interesting…

ChatGPT’s Training Program

When we asked ChatGPT to write us a training program, it gave us a pretty simple training program that wasn’t inherently bad. We did however notice that it lacked a few key components.

We asked: “Hey ChatGPT! Can you please write me a training program for 2 days a week in a table format?” Check out the program it laid out for us below. It included similar exercises to what we would incorporate in our client’s training programs at Ivy Training, however we can see a few red flags (which like most red flags, may not be obvious at first glance).

Red Flags 🚩

Before we get into the nitty-gritty details of ChatGPT’s program, we’d like to say that getting yourself to the gym and following any training program is better than doing nothing. However, here are a few red flags that we noticed when reading over it.

1. There is no reference to intensity.

We covered The Four Pillars of Fitness recently, and intensity is one of those important pillars. Any decent training program will specify intensity by using RPE (Rate of Perceived Exertion), RIR (Repetitions in Reserve) or a percentage of e1RM (estimated 1 Repetition Max). ChatGPT doesn’t reference anything to do with intensity, so how are we meant to know how hard we’re meant to push these sets? What loads should we select? Because the level of effort is left undefined, we’re left in the dark about weight selection. At Ivy Training, we primarily use RPE to prescribe effort and gauge how hard a set is for our clients. We don’t want our clients working to failure, but we also don’t want it to be a walk in the park. Our aim is to select an intensity that is challenging enough to be stimulating, without accumulating too much fatigue.

2. There is no variation in the sets and reps.

The next red flag that we noticed is that each exercise is 3 sets of 8-10 reps. Whilst there is nothing inherently wrong with sets of 8-10, it doesn’t necessarily allow for higher intensities. This is because the number of reps is inversely related to the training intensity or load. We would also argue that most individuals would benefit from using a variety of different rep ranges. Lower reps will increase the training intensity and expose your muscles to high levels of force, which will result in unique adaptations that improve stiffness under heavy loads (making someone stronger). Conversely, higher reps are useful for accumulating enough training volume in a time-efficient manner and secondly, improves local muscular endurance.

3. It doesn’t tell us how long we need to rest for.

Depending individual factors, rest time will vary program to program and exercise to exercise. ChatGPT has failed to establish any rest period, so how are we meant to know how long to rest? Is 30 seconds enough? Is 5 minutes too much? It all depends on the context! Generally speaking, we like to keep our rest times between 30 seconds and 3 minutes. This is quite a difference, but as a guide we will typically use shorter rest periods (30-60 seconds) for smaller isolated/accessory exercises and longer rest periods (2-3 minutes) for larger barbell movements. The reason we allow longer rest periods for barbell movements is due to them recruiting larger muscle groups, and therefore requiring more recovery between sets. It takes us a bit longer to feel “ready” (physically and mentally) to hit a heavy set of deadlifts than it does for a dumbbell bicep curl.

4. There is no description on warming up to your working sets.

This was the most problematic for us. We would describe warming up as preparing the body for the physical demands of the task at hand, and priming the nervous system to best perform that task at hand. We’d argue that anything else is either superfluous, fatiguing or unhelpful. Whilst there is no harm in doing “light cardio exercise for 5-10 minutes” as ChatGPT outlined, that type of physical activity isn’t going to prime you for your Barbell Squats that you have following. At Ivy Training, we like to start with body weight or the empty barbell, before increasing the load incrementally until you get to your target weight. We have another fantastic blog on Everything You Need To Know About Warming Up if you’re eager to learn more.

5. Does the cool down serve any purpose?

At the end of the program, ChatGPT prescribed 5-10 minutes of “Any stretching exercises for major muscle groups”. Not only do we find this a bit vague, but we’d also question the purpose. If we dig into the latest research on stretching, there is not a sufficient amount of evidence to support the implementation of stretching to prevent injury or improve joint mobility. We’d say skip the stretch and add in some more work for that 5-10 minutes, whether that be an extra set or an extra exercise.

Conclusion

As we mentioned earlier, it’s better to have some structure to your workout than none at all. However, ChatGPT has oversimplified the process of writing a training program for an individual. It doesn’t take into account the important red flags outlined throughout this blog, and overall we’d argue that it doesn’t structure the training week as well as what what we (or other personal trainers) could. It didn’t ask us about our goals or capabilities, or whether we had any injuries that needed to be accounted for. ChatGPT definitely has a place in this world (much to Jake’s dismay), but sometimes you can’t beat human interaction and an individualised approach.

Resources

  1. https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD007455.pub3/full?highlightAbstract=stretching&highlightAbstract=stretch
  2. https://www.painscience.com/articles/stretching.php#rr
  3. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8120977/
  4. https://pubmed.ncbi.nlm.nih.gov/29470825/
  5. https://pubmed.ncbi.nlm.nih.gov/29324578/
  6. https://pubmed.ncbi.nlm.nih.gov/31260419/