Table 1. ChatGPT performance by dataset (Orthobullets vs. 2022 Orthopedic In-Training Exam [OITE] vs. 2021 OITE).
Source | Orthobullets (n = 239) | 2022 OITE (n = 124) | 2021 OITE (n = 109) | Total (n = 472) |
Correct | 131 (54.8%) | 74 (59.7%) | 59 (54.1%) | 264 (55.9%) |
Incorrect | 108 (45.2%) | 50 (40.3%) | 50 (45.9%) | 208 (44.1%) |