Table 2.
Consensus-based accuracy ratings of LLM-Chatbot responses to ocular symptoms-related questions
| Domaina | No. of Questions | ChatGPT-3.5, n (%) |
ChatGPT-4.0, n (%) |
Google bard, n (%) |
||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Poor | Borderline | Good | Poor | Borderline | Good | Poor | Borderline | Good | ||
| Visual Disturbance Symptomsb | 18 | 1 (5.6) | 8 (44.4) | 9 (50.0) | 0 (0) | 4 (22.2) | 14 (77.8) | 3 (16.7) | 11 (61.1) | 4 (22.2) |
| Other Ocular Symptomsc | 19 | 1 (5.3) | 5 (26.3) | 13 (68.4) | 0 (0) | 0 (0) | 19 (100) | 2 (10.5) | 6 (31.6) | 11 (57.9) |
a
Based on majority consensus across the three graders.
b
Includes acute vision loss, chronic vision loss, diplopia, metamorphopsia, photopsia, myodesopsia, visual field defect, photophobia, and glare sensitivity/haloes.
c
Includes red eye, painful eye, pruritus, pressure sensation, epiphora, ptosis, blepharospasm, foreign body sensation, trauma, and asthenopia