Artificial Intelligence (or AI) has been in the headlines and commercials a lot in recent months. As a business owner and coach, I am looking to use AI as a tool to enhance my coaching. For example, the app I use for running form analyses, Ochy, uses AI. However, as with all technologies there are pros and cons. The internet has changed our life forever, but it certainly hasn’t been all good. It can be a platform for misinformation, the spread of hate, and detrimental impacts to our mental health.
With both the pros and cons in mind, I asked Meta AI and Gemini AI to create a 16-week marathon plan. I was curious how AI generated training plans compare to static training plans you would find on any number of apps and websites (80/20, TrainingPeaks, Hal Higdon app, etc.).
Here is a summary on what Meta AI and Gemini AI created based on the prompt, ’16-wek training plan for a marathon’.
Gemini AI marathon training plan
Google’s Gemini started with a brief introduction stating that this plan was good for beginner to intermediate runners and that it was important to adjust it based on your individual fitness and goals. They went on to define terms including easy run, long, run, rest, and strength training.
The whole training plan neatly fits into one table:
Main Issues With Gemini AI Training Plan:
- Doesn’t guide you with how to create the 16-week outline. Only that you should gradually increase the long run from 6 miles to 18 – 20 miles
- Does not mention that you need recovery weeks as part of a training plan
- Does not give you any guidance with setting up training zones and how to determine what an easy or interval pace should be
- No guidelines for what an interval session should look like (pace, work to rest ratio, time at intensity).
- The only run that changes week to week is the long run. Otherwise, the plan looks identical every week.
Main Takeaway:
Gemini AI gets a point for suggesting 1 – 2 rest days per week and another point for suggesting a slow build up to 18 – 20 miles. Besides that, the plan is really not good. If you can’t afford to hire a personal running coach, you will get a far better product by spending $30 – $50 on a static training plan designed by a professional running coach.
Meta AI training plan
Meta’s AI did not have an introduction with definitions. They listed the training plan broken down into four week phases. There are so many red flags with this plan I will simply just have to get right down to the negatives.
Main Issues With Meta AI Training Plan:
- The names of the phases are vague and aren’t in line with sports science research. The base phase is a bit practical but then it’s followed by building endurance. Base building = building endurance. The third phase is building strength and endurance. It’s not clear why they only suggest starting strength training in week 9 of the training plan. You will be a grumpy athlete indeed if you are contending with delayed onset muscle soreness from a new strength routine as well as completing your longest runs of the training plan. It also isn’t clear why the taper phase happens four and three weeks out from the race instead of two and one week out from a race which is the standard practice. They are recommending a four week taper which is very long for a beginner to intermediate runner that doesn’t have much mileage in the last several months to taper from.The longest run in the training plan is only 10 miles. A 4-week taper plus only completing a 10 mile long run as your longest run! That 26.2 miles is going to hurt not to mention could potentially be dangerous to your health without adequate preparation.
- There is no guidance with regards to paces you should run or even what effort runs should be completed at. Training load includes both volume and intensity and this plan completely ignores intensity.
- They don’t provide information on the importance of recovery weeks.
Main Takeaway:
They get one point for at least providing information for the 16-weeks. Besides that, this plan is disastrous!