From Quality to Quantity: AI-Driven Publication Inflation and Its Impact on Dermatology Applications

Abstract

Introduction: Following the public release of ChatGPT and similar large language models (LLMs) in late 2022, AI-assisted writing has become increasingly accessible. The use of AI in academic writing is particularly concerning. In dermatology, where publication numbers per residency applicant have steadily risen in recent cycles, concerns exist about research authenticity and the weight of publications in applicant evaluations. Methods: To assess the presence of AI in dermatology publications, we analyzed 200 case reports from the Journal of the American Academy of Dermatology (JAAD) using ZeroGPT, an AI detection tool. Reports were selected chronologically to maintain objectivity: 100 from 2022, before AI became public, and 100 from late 2024-early 2025, after AI became readily available. Publications were further stratified based on the first author's academic status as a medical student (bachelor’s degree with medical school affiliation) or physician (MD/DO/MBBS). Mann-Whitney U Tests compared AI content detected in reports from 2022 and more recent years, and between author groups. Results: AI-identified content significantly increased after 2022, rising from 18.8 ± 12.6 to 38.6 ± 18.6 (p < 0.0001). No significant difference was found between medical student and physician first authors (p = 0.47). Conclusion: The post-2022 surge in AI-generated content, consistent across author levels, highlights AI’s widespread adoption. Given the recent rise in applicant publication numbers, residency programs should reassess the significance of publication numbers in their evaluations. Future research should investigate specific ways applicants are utilizing AI in research.

Ähnliche Arbeiten