A reproducible statistical evaluation framework for large-sample assessment of AI-generated medical references: cross-platform application of the reference hallucination score
Nurmuhammet Taş, Yakup Erden, Mustafa Hüseyin Temel, Fatih Bağcıer · 2026
Read A reproducible statistical evaluation framework for large-sample assessment of AI-generated medical references: cross-platform application of the reference hallucination score by Nurmuhammet Taş, Yakup Erden, Mustafa Hüseyin Temel, Fatih Bağcıer (2026) 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/PMC13523266.txt.
Description
Nurmuhammet Taş and colleagues build a scoring framework for checking source lists generated by chatbots in a 2026 cross-platform study. Three chatbots were prompted under thirty medical headings in two output formats, producing more than three thousand entries. Each entry was scored for whether it could be found and verified, for record accuracy, for identifier validity, and for topical relevance.
Total scores differed clearly between models, yet the overall effect size was very small, meaning the practical gap was modest. Longer outputs in review format showed somewhat weaker record stability, with differences shrinking after correction. The authors position their score less as a league table and more as a reusable, instrument-neutral framework for large-scale trust studies of generated source lists.
Catalog record
| CREATOR | Nurmuhammet Taş, Yakup Erden, Mustafa Hüseyin Temel, Fatih Bağcıer |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 47967 |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |