ChatGPT in head and neck scientific writing: A precautionary anecdote.
Abstract
Purpose To evaluate the accuracy of ChatGPT references in scientific writing relevant to head and neck surgery. Materials and methods Five commonly researched keywords relevant to head and neck surgery were selected (osteoradionecrosis of the jaws, oral cancer, adjuvant therapy for oral cancer, TORS, and free flap reconstruction in oral cancer). The AI chatbot was then asked to provide ten complete citations for each of the keywords. Two independent authors reviewed the results for accuracy and assigned each article a numerical score based on pre-selected criteria. Results Among 50 total references provided by ChatGPT, only five (10 %) were found to have the correct title, journal, authors, year of publication, and DOI. Merely 14 % of the presented references had correct DOI. References regarding free flap reconstruction for oral cancer were the least accurate from all the five categories, with no correct DOI. Complete inter-rater agreement was noted while evaluating the citations. Conclusion Only 10 % of the articles provided by ChatGPT, relevant to head and neck surgery, were correct. A high degree of academic hallucination was noted.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Only 10% of references generated by ChatGPT in medical writing were fully correct, while DOI hallucinations occurred in the majority of cases.
“Wu and Dang (2023) further demonstrated that only 10% of references generated by ChatGPT in the context of medical writing were fully correct, while DOI hallucinations occurred in the majority of cases.”
From Training Students to Identify and Correct Fabricated References in ChatGPT-Generated Literature Reviews (2025) · cites this paper as Wu and Dang (2023)
AI tools frequently output inconsistently formatted references.
“Unlike human writers who may gradually learn formatting rules, AI tools frequently output inconsistently formatted references (Wu & Dang, 2023).”
From Training Students to Identify and Correct Fabricated References in ChatGPT-Generated Literature Reviews (2025) · cites this paper as Wu & Dang, 2023
The citation lists haven’t been fetched yet
References 10 · Cited by 39
- Authors
- Wu, R. T., & Dang, R. R.
- Affiliations
- Chang Gung Memorial Hospital (TW)
- Boston Medical Center (US)
- Corresponding author
- Dang, Rushil R.
- Venue
- American Journal of Otolaryngology
- Publisher
- Elsevier BV
- Published
- 2023-11
- Volume
- 44
- Issue
- 6
- Pages
- 103980
- ISSN
- 0196-0709 (ISSN-L)
- OpenAlex
- W4383376748
- Cited by
- 39
- References
- 10
- Citations by year
- 2026: 6 · 2025: 13 · 2024: 16 · 2023: 4
- FWCI
- 1.58
- Type
- journal article
- Language
- en
- Open access
- closed
- Licence
- Locations
- American Journal of Otolaryngology (publishedVersion)
- PubMed (publishedVersion)
- Topics
- Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
- Social Sciences › Arts and Humanities › History and Philosophy of Science › Academic Writing and Publishing
- Social Sciences › Safety Research › Academic integrity and plagiarism
- Keywords
- ChatGPT, large language models, AI-generated references, citation accuracy, head and neck surgery, oral cancer, free flap reconstruction, transoral robotic surgery, osteoradionecrosis of the jaw, adjuvant therapy
- MeSH terms
- Combined Modality Therapy, Head, Humans, Mouth Neoplasms, Neck, Writing
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
