To evaluate the appropriateness of responses generated by ChatGPT-5 for common acne-related questions.
Approach:
Study Design: A cross-sectional observational study using 25 acne-related questions categorized into skin care routine, nutrition and lifestyle, diagnosis and general information, and treatment and medications.
Response Evaluation: Each question was submitted to ChatGPT-5 in three independent sessions, generating 75 responses, which were classified by three dermatologists as appropriate or inappropriate.
Key Findings:
65% of responses were classified as appropriate.
Appropriateness was highest for skin care (83%) and lowest for diagnosis and general information (33%).
Misleading information was the most common reason for inappropriate ratings.
Limitations:
The study was limited to acne, affecting generalizability to other dermatologic conditions.
Binary classification may not capture intermediate degrees of appropriateness.
Low inter-rater agreement among dermatologists raises concerns about the consistency of evaluations.
Analyses by treatment timing, cumulative dose, and stunting supported the overall finding, though early-initiation and female subgroup data were limited.
