Exploring the Ability of ChatGPT to Act as a Research Aid in Otolaryngology.

Abstract

Recently artificial intelligence (AI) platforms have developed at a rapid pace. To date no studies have explored AI platform ChatGPT's ability to serve as an aid in research in the field of otolaryngology. The objective of our study is to evaluate the ability of ChatGPT to generate unique research ideas relevant to Otolaryngology. ChatGPT was tasked to generate novel research project ideas. Seven categories for all otolaryngology subspecialties were created: general otolaryngology, facial plastics and reconstructive surgery, rhinology and skull base surgery, pediatrics, head and neck oncology, laryngology, and otology/neurotology. Within each of the subspecialties, ChatGPT was prompted to provide research ideas for two specific research topics in order to gauge ChatGPT's ability to explore specific domains. Ten prompts were entered for general otolaryngology, and five prompts were entered for each subspecialty and subspecialty topic. ChatGPT was subsequently prompted with the same query to generate systematic review ideas for each category. Ideas were deemed novel if there were no similar systematic reviews retrieved during the literature review on PubMed, Scopus, or Web of Science. Ideas were graded for Clinical Relevance on a scale of 0 to 5, 5 being considered highly relevant/would contribute greatly to patient care and 0 being considered not relevant/not beneficial to patient care. Two reviewers graded each response. Out of 100 systematic review ideas generated by ChatGPT, we found that the artificial intelligence platform was incapable of creating unique systematic review research ideas. However, the ideas that it did generate were largely feasible and clinically relevant. Future studies should investigate the ability of ChatGPT to generate research inquiries with non-systematic review methodologies and in more specific otolaryngology topics.