ChatGPT Generated Training Plans for Runners are not Rated Optimal by Coaching Experts, but Increase in Quality with Additional Input Information

Abstract

ChatGPT may be used by runners to generate training plans to enhance performance or health aspects. However, the quality of ChatGPT generated training plans based on different input information is unknown. The objective of the study was to evaluate ChatGPT-generated six-week training plans for runners based on different input information granularity. Three training plans were generated by ChatGPT using different input information granularity. 22 quality criteria for training plans were drawn from the literature and used to evaluate training plans by coaching experts on a 1-5 Likert Scale. A Friedmann test assessed significant differences in quality between training plans. For training plans 1, 2 and 3, a median rating of <3 was given 19, 11, and 1 times, a median rating of 3 was given 3, 5, and 8 times and a median rating of >3 was given 0, 6, 13 times, respectively. Training plan 1 received significantly lower ratings compared to training plan 2 for 3 criteria, and 15 times significantly lower ratings compared to training plan 3 (p < 0.05). Training plan 2 received significantly lower ratings (p < 0.05) compared to plan 3 for 9 criteria. ChatGPT generated plans are ranked sub-optimally by coaching experts, although the quality increases when more input information are provided. An understanding of aspects relevant to programming distance running training is important, and we advise avoiding the use of ChatGPT generated training plans without an expert coach’s feedback.