Comparing Ophthalmologist and Artificial Intelligence Chatbot Responses to Patient Questions
Bibliographic record
Abstract
Purpose: We evaluated the ability of ChatGPT, an Artificial Intelligence (AI) Chatbot, to respond to patient eye health queries. Methods: A retrospective, cross-sectional analysis of eye health questions and physician responses posted on the American Academy of Ophthalmology (AAO) "Ask an Ophthalmologist" forum was performed on a random sample from January 2016 to December 2022. We compared board-certified ophthalmologists' responses to ChatGPT (version GPT-4o, OpenAI) responses in September 2024. Primary outcomes included ophthalmologist-rated accuracy of ChatGPT and AAO responses using a 7-point Likert scale, as well as ophthalmologists' preferences between the two responses. Secondary outcomes assessed differences in readability, empathy, and response length between ChatGPT and ophthalmologists. Results: A random sample 250 questions and responses from 41 board-certified ophthalmologists were evaluated. ChatGPT and AAO responses had similar mean accuracy ratings (5.8 [SD=1.1] vs 5.5 [SD=1.1], p=0.07). Evaluators preferred ChatGPT over physician responses in half (49.5%) the cases. Ophthalmologist responses were easier to understand, with a lower mean Flesch-Kincaid Grade Level (Grade 11.0 [SD=2.7] vs Grade 12.7 [SD=1.9], p<0.001). Ophthalmologist responses were also significantly shorter than ChatGPT responses (80.6 [SD=56.4] words vs (337.8 [SD=141.6] words, p<0.001). Empathy ratings did not differ significantly between ChatGPT and ophthalmologists (4.4 [SD=0.6] vs 4.4 [SD=0.6], p=0.5). Conclusion: Our findings suggest that Chatbot responses were as frequently preferred as physician responses, rated with higher accuracy, and demonstrated comparable empathy in addressing online patient eye health queries. AI chatbots may assist in drafting initial responses to patient concerns, potentially improving efficiency and reducing physician workload.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.016 | 0.095 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.000 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".