Uydurma mı gerçek mi? Kas-iskelet sistemi konularında yapay zeka tarafından üretilen bilimsel yazılarda etik sorunlar ve atıf halüsinasyonları

Abstract

Amaç: ChatGPT gibi büyük dil modelleri (LLM'ler), akademik ve klinik yazımda giderek daha fazla kullanılmaktadır. Bu araçlar tutarlı ve alanına özgü metinler üretebilse de, otomatik olarak oluşturulan referansların doğruluğuna ilişkin endişeler devam etmektedir. Güncel kanıtlara büyük ölçüde dayanan kas-iskelet sistemi rehabilitasyonu alanında, atıfların güvenilirliği özellikle kritik öneme sahiptir. Ancak yapay zekâ tarafından üretilen bilimsel içeriklerde atıf doğruluğunu sistematik olarak değerlendiren çalışmalar yetersizdir. Bu çalışmanın amacı, kas-iskelet sistemi rehabilitasyonu konularında oluşturulan promptlara yanıt olarak ChatGPT (GPT-4) tarafından üretilen bilimsel metinlerdeki referansların doğruluğunu değerlendirmek ve yapılandırılmış bir doğrulama sürecinin referans doğruluğunu artırıp artırmadığını belirlemektir. Yöntem: ChatGPT’den manuel terapi, ön çapraz bağ (ACL) rekonstrüksiyonu, bel ağrısı ve rotator manşet onarımı konularında, her biri 10 DOI içeren dört bilimsel paragraf üretmesi istendi. Toplam 40 referans, 3 puanlık bir sistemle değerlendirildi: 0 = uydurma, 1 = kısmen doğru, 2 = tamamen doğru. İlk değerlendirmeden sonra ChatGPT'den referansları doğrulaması ve düzeltmesi istendi. İlk ve son durumdaki puanlar betimsel olarak karşılaştırıldı. Bulgular: İlk üretimde referansların yalnızca %7,5’i tamamen doğruydu, %42,5’i ise tamamen uydurmaydı. Geri kalan %50’si kısmen doğruydu. Doğrulama sürecinden sonra tamamen doğru referans oranı %77,5’e yükseldi. En yaygın hatalar geçersiz DOI’ler, uydurma makale başlıkları ve yanlış eşleşmiş metadata içeriğiydi. Sonuç: ChatGPT tutarlı bilimsel metinler üretebilse de, referansları sıklıkla hatalı veya uydurmadır. Üretim sonrası doğrulama, referans doğruluğunu önemli ölçüde artırmaktadır. Yapay zekâ araçlarının akademik ve klinik kas-iskelet sistemi bağlamlarında kullanımı sırasında, özellikle atıf geçerliliği açısından dikkatli olunmalıdır.

Keywords

• ChatGPT
• Yapay zeka
• Muskuloskeletal rehabilitasyon
• bilimsel yazı
• referans doğruluğu
• atıf halüsinasyonu

Fabricated or accurate? Ethical concerns and citation hallucination in aI-generated scientific writing on musculoskeletal topics

Abstract

Aims: Large language models (LLMs) such as ChatGPT are increasingly used in academic and clinical writing. While these tools can generate coherent and domain-specific text, concerns persist regarding the accuracy of their automatically generated references. In musculoskeletal rehabilitation—a field heavily reliant on current evidence—the reliability of citations is especially critical. Yet, systematic evaluations of citation accuracy in AI-generated scientific content are lacking. To evaluate the reference accuracy of scientific texts generated by ChatGPT (GPT-4) in response to musculoskeletal rehabilitation prompts, and to determine whether reference accuracy improves following structured post-generation verification. Methods: ChatGPT was prompted to generate four scientific paragraphs on musculoskeletal rehabilitation topics (manual therapy, ACL reconstruction, low back pain, and rotator cuff repair), each including 10 references with DOIs. A total of 40 references were analyzed using a 3-point scoring system (0=fabricated, 1=partially correct, 2=fully accurate), which was used to assess citation quality. After initial evaluation, ChatGPT was asked to verify and revise its references. Scores before and after this step were compared descriptively and with Wilcoxon signed-rank tests to assess statistical significance, and effect sizes (r) were calculated to estimate the magnitude of improvement. Results: Only 7.5% of references were fully accurate in the initial generation, while 42.5% were completely fabricated. The remaining 50% were partially correct. After verification, the proportion of fully accurate references rose to 77.5%. Wilcoxon signed-rank testing confirmed a statistically significant improvement in accuracy across all prompts (W=561.0, p<0.001, r=0.60). The most common errors included invalid DOIs, fabricated article titles, and mismatched metadata. Conclusion: ChatGPT can generate coherent scientific content, but its initial references are frequently inaccurate or fabricated. Structured post-generation verification significantly improves reference accuracy, as confirmed by statistical testing. These findings suggest that LLMs may be integrated as drafting tools in academic and clinical musculoskeletal contexts, but only when accompanied by strict human-led verification of citations.

Keywords

• ChatGPT
• artificial intelligence
• musculoskeletal rehabilitation
• scientific writing
• reference accuracy
• hallucinated citations

References

1. Mondal H, Mondal S. ChatGPT in academic writing: maximizing its benefits and minimizing the risks. Indian J Ophthalmol. 2023;71(12): 3600-3606. doi:10.4103/IJO.IJO_718_23
2. Bom H-SH. Exploring the opportunities and challenges of ChatGPT in academic writing: a roundtable discussion. Nucl Med Mol Imaging. 2023;57(4):165-1677. doi:10.1007/s13139-023-00809-2
3. Jarrah AM, Wardat Y, Fidalgo P. Using ChatGPT in academic writing is (not) a form of plagiarism: what does the literature say. Online J Commun Media Technol. 2023;13(4):e202346. doi:10.30935/ojcmt/13572
4. Gruda D. Three ways ChatGPT helps me in my academic writing. Nature. 2024;10:1-6. doi:10.1038/d41586-024-01042-3
5. Švab I, Klemenc-Ketiš Z, Zupanič S. New challenges in scientific publications: referencing, artificial intelligence and ChatGPT. Slov J Public Health. 2023;62(3):109-112. doi:10.2478/sjph-2023-0015
6. Jan R. Examining the reliability of ChatGPT: identifying retracted scientific literature and ensuring accurate citations and references. In: Impacts of generative ai on the future of research and education. Hershey, PA: IGI Global; 2025.
7. Frosolini A, Gennaro P, Cascino F, Gabriele G. In reference to “role of Chat GPT in public health”, to highlight the AI’s incorrect reference generation. Ann Biomed Eng. 2023;51(10):2120-2122. doi:10.1007/s 10439-023-03248-4
8. Cohen F, Vallimont J, Gelfand AA. Caution regarding fabricated citations from artificial intelligence. Headache. 2024;64(1):133-135. doi: 10.1111/head.14649

9. Borji A. A categorical archive of ChatGPT failures. arXiv. 2023;2302. 03494. doi:10.48550/arXiv.2302.03494
10. Bai Z, Wang P, Xiao T, et al. Hallucination of multimodal large language models: a survey. arXiv. 2024;2404.18930. doi:10.48550/arXiv.2404.18930
11. Metze K, Morandin-Reis RC, Lorand-Metze I, Florindo JB. Bibliographic research with ChatGPT may be misleading: the problem of hallucination. J Pediatr Surg. 2024;59(1):158-159. doi:10.1016/j.jpedsurg.2023.08.018
12. Aljamaan F, Temsah M-H, Altamimi I, et al. Reference hallucination score for medical artificial intelligence chatbots: development and usability study. JMIR Med Inform. 2024;12(1):e54345. doi:10.2196/54345
13. Mohammad B, Supti T, Alzubaidi M, et al. The pros and cons of using ChatGPT in medical education: a scoping review. In: Healthcare transformation with informatics and artificial intelligence. Volume 305. 2023.
14. Gao CA, Howard FM, Markov NS, et al. Comparing scientific abstracts generated by ChatGPT to original abstracts using an artificial intelligence output detector, plagiarism detector, and blinded human reviewers. bioRxiv. 2022;12(23):521610. doi:10.1101/2022.12.23.521610
15. Sharun K, Banu SA, Pawde AM, et al. ChatGPT and artificial hallucinations in stem cell research: assessing the accuracy of generated references–a preliminary study. Ann Med Surg (Lond). 2023;85(10):5275-5278. doi:10.1097/MS9.0000000000001228
16. Safran E, Yildirim S. A cross-sectional study on ChatGPT’s alignment with clinical practice guidelines in musculoskeletal rehabilitation. BMC Musculoskelet Disord. 2025;26(1):411. doi:10.1186/s12891-025-08650-8
17. Lancaster T. Artificial intelligence, text generation tools and ChatGPT–does digital watermarking offer a solution? Int J Educ Integr. 2023;19(1):10.
18. Alkaissi H, McFarlane SI. Artificial hallucinations in ChatGPT: implications in scientific writing. Cureus. 2023;15(2):e35074. doi:10. 7759/cureus.35179
19. Thorp HH. ChatGPT is fun, but not an author. Science. 2023;379(6630): 313. doi:10.1126/science.adg7879
20. Gravel J, D’Amours-Gravel M, Osmanlliu E. Learning to fake it: limited responses and fabricated references provided by ChatGPT for medical questions. Mayo Clin Proc Digit Health. 2023;1(3):226-234. doi:10.1016/j.mcpdig.2023.05.004
21. Orduña-Malea E, Cabezas-Clavijo Á. ChatGPT and the potential growing of ghost bibliographic references. Scientometrics. 2023;128(9): 5351-5365. doi:10.1007/s11192-023-04804-4
22. Lechien JR, Briganti G, Vaira LA. Accuracy of ChatGPT-3.5 and -4 in providing scientific references in otolaryngology–head and neck surgery. Eur Arch Otorhinolaryngol. 2024;281(4):2159-2165. doi:10.1007/s00405-023-08441-8
23. Sanchez-Ramos L, Lin L, Romero R. Beware of references when using ChatGPT as a source of information to write scientific articles. Am J Obstet Gynecol. 2023;229(3):356. doi:10.1016/j.ajog.2023.04.004
24. Sebo P. How accurate are the references generated by ChatGPT in internal medicine? Intern Emerg Med. 2024;19(1):247-249. doi:10.1007/s11739-023-03484-5
25. Besançon L, Cabanac G, Labbé C, Magazinov A. Sneaked references: fabricated reference metadata distort citation counts. J Assoc Inf Sci Technol. 2024;75(12):1368-1379. doi:10.48550/arXiv.2310.02192
26. Walters WH, Wilder EI. Fabrication and errors in the bibliographic citations generated by ChatGPT. Sci Rep. 2023;13(1):14045. doi:10.1038/s41598-023-41032-5
27. Graf EM, McKinney JA, Dye AB, Lin L, Sanchez-Ramos L. Exploring the limits of artificial intelligence for referencing scientific articles. Am J Perinatol. 2024;41(15):2072-2081. doi:10.1055/s-0044-1786033
28. Altmäe S, Sola-Leyva A, Salumets A. Artificial intelligence in scientific writing: a friend or a foe? Reprod Biomed Online. 2023;47(1):3-9. doi:10. 1016/j.rbmo.2023.04.009
29. Watson AP. Hallucinated citation analysis: delving into student-submitted AI-generated sources at the University of Mississippi. Ser Libr. 2024;85(5-6):172-180. doi:10.1080/0361526X.2024.2433640
30. Frosolini A, Franz L, Benedetti S, et al. Assessing the accuracy of ChatGPT references in head and neck and ENT disciplines. Eur Arch Otorhinolaryngol. 2023;280(11):5129-5133. doi:10.1007/s00405-023-08205-4
31. Giray L. ChatGPT references unveiled: distinguishing the reliable from the fake. Internet Ref Serv Q. 2024;28(1):9-18. doi:10.1080/10875301.2023. 2265369
32. Athaluri SA, Manthena SV, Kesapragada VKM, Yarlagadda V, Dave T, Duddumpudi RTS. Exploring the boundaries of reality: investigating the phenomenon of artificial intelligence hallucination in scientific writing through ChatGPT references. Cureus. 2023;15(4):e37432. doi:10. 7759/cureus.37432
33. Bhattacharyya M, Miller VM, Bhattacharyya D, Miller LE, Miller V. High rates of fabricated and inaccurate references in ChatGPT-generated medical content. Cureus. 2023;15(5):e39238. doi:10.7759/cureus.39238