This study aims to examine the validity of prescribed AI-generated exercise interventions for general health and weight loss. This study developed an AI-generated exercise program using the Delphi method. The Delphi process involved administering a structured questionnaire within the relevant domain, whereby a panel of experts evaluated the appropriateness of the program. To assess validity, three established measures were employed: the Content Validity Ratio (CVR), the Content Validity Index (CVI), and the Impact Score (IS). In addition, the Fleiss’ Kappa coefficient (κ) was calculated to determine the degree of agreement (reliability) among the experts. All statistical analyses were conducted using SPSS software version 27 and Microsoft Excel.
The IS indicates that all exercises achieved the required level of validity. Also, based on the CVI and CVR, all exercises demonstrated acceptable content validity for general health and weight loss. However, the negative kappa value suggests poor agreement among the experts for the mentioned variables.
The findings indicate that while platforms such as ChatGPT-4o can produce broadly appropriate exercise content, expert agreement on the validity of these protocols was very low. Accordingly, AI should be viewed as a supplementary tool that can assist in designing individualized health and weight management programs.