The Performance of GPT-3.5, GPT-4, and Bard on the Japanese National Dentist Examination: A Comparison Study
Keiichi Ohta, Satomi Ohta · 2023
Read The Performance of GPT-3.5, GPT-4, and Bard on the Japanese National Dentist Examination: A Comparison Study by Keiichi Ohta, Satomi Ohta (2023) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-10782219.txt.
Description
Keiichi Ohta and Satomi Ohta sit three chatbots for Japan's national dentist examination and grade them against the human pass mark. Their 2023 comparison gives 185 questions from that year's exam to GPT-3.5, GPT-4, and Google's Bard, broken down by question type and subject.
GPT-4 answered about seventy-four percent correctly overall, Bard about sixty-seven, and the older model about fifty-two. On the essential-questions section, though, only Bard cleared the eighty percent passing bar; GPT-4 fell short despite its lead. All three models sagged on professional dentistry questions compared with other categories. The authors read the results as a capability map, not a coronation: general reasoning is strong, specialist dental knowledge is not yet exam-grade. For regulators and educators, the nuance matters more than the headline. A model can top the class and still fail the section that licenses it.
Catalog record
| CREATOR | Keiichi Ohta, Satomi Ohta |
|---|---|
| YEAR | 2023 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 15 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |