Comparative utility analysis of Chordoma search information between ChatGPT vs. Google Web

Abstract

Objective

This study compares the utility of ChatGPT and Google searches in obtaining information about chordoma, a spine pathology. It is hypothesized that ChatGPT will provide a broader range of questions and more reliable sources due to its adaptive learning.

Methods

A Google and ChatGPT search for "chordoma" was performed, recording the first 10 FAQs and their sources. Responses to the 10 most common FAQs were collected and classified using the Rothwell scheme.

Results

There were 3 of 10 questions (30 %) similar amongst FAQs provided by a Google Web and ChatGPT search for the term “chordoma.” The most abundant Rothwell question category from Google, 4 of 10 questions (40 %), was “technical details.” The remaining questions were as follows: timeline of recovery (20 %), indications/management (20 %), and risks/complications (20 %). Pertaining to ChatGPT, the most abundant question classification was “specific activities,” 3 out of 10 questions (30 %). Remaining distribution was technical details (20 %), indications/management (20 %), risks/complications (20 %), and timeline of recovery (10 %). Regarding Google, 3 of the 10 questions asked were associated with a response from a commercial website, unlike 1 out of 10 for ChatGPT. In addition, ChatGPT predominantly utilized government sources (70 %), most frequently PubMed. Google's most abundant source type was academic (50 %). All numerical questions (100 %) had varied answers between Google and ChatGPT.

Conclusion

Differences exist between chordoma-related information from ChatGPT and Google. ChatGPT relies more on government sources, making it a useful adjunct tool for patients seeking spine pathology information. Further research is needed to assess its clinical applicability.