Ria Resti Fauziah, Ari Metalin Ika Puspita, Ivo Yuliana, Fiena Saadatul Ummah, Siti Mufarochah, Erfan Ramadhani
This letter discusses findings from a recent study comparing AI-generated and humanwritten neurosurgery articles. The study reveals that AI-generated articles exhibit higher readability scores (Lix: 35 vs. 26, Flesch-Kincaid: 10 vs. 8) but may lack depth in analysis. Evaluators could correctly identify AI authorship with 61 % accuracy, and preferences were nearly even between AI-generated (47 %) and human-written (44 %) articles. While AI improves accessibility and efficiency in academic writing, its limitations in clinical experience, originality, and nuanced analysis highlight the need for human oversight. The integration of AI should be as a complementary tool rather than a replacement for human expertise. Future research should focus on refining AI's analytical capabilities and ensuring ethical use in scientific publishing. © 2025 Elsevier Ltd
Institut Al Azhar Menganti Gresik, Jl. Raya Menganti Krajan 474 Menganti Gresik, Indonesia; Universitas Negeri Surabaya Jl. Lidah Wetan, Surabaya, Indonesia; Department Guidance and Counseling, Universitas PGRI Palembang, Indonesia Jln. Jend. A. Yani, Lr Gotong royong 9/10 Ulu, Palembang, Indonesia
