Parent Questions About Childhood Hearing Loss: An Evaluation of ChatGPT Response Accuracy, Completeness, and Repeatability

Purpose: We aimed to assess the responses to parent questions about childhood hearing loss that were generated by two generative artificial intelligence (AI) chatbots, ChatGPT-4o mini and ChatGPT Professional (Pro). Method: Sixty-four questions grouped into three categories (hearing loss, hearing aids, and family) were queried in both platforms and rated for accuracy, completeness, readability, and repeatability. Paired-samples t tests were calculated to determine differences in responses between platforms. Results: ChatGPT Pro was judged as more accurate for hearing aids and more complete for questions related to hearing loss and hearing aids compared to ChatGPT-4o mini, and ChatGPT-4o mini responses were judged as more complete compared to ChatGPT Pro for questions related to family (e.g., emotions, talking to others). Readability scores revealed that responses were generally written in a complex manner. Conclusions: Responses generated on both platforms were generally judged as accurate and complete, suggesting that AI could contribute to supporting parent education needs. Further research is needed to explore strategies for improving the accuracy and completeness of AI-generated responses, including the collection of supplemental information from parent users to guide and/or prompt responses.