ChatGPT in Neurosurgical Practice and Education: Benefits, Risks, and Challenges

This systematic review evaluates the integration of ChatGPT into neurosurgical practice and education by analyzing 23 peer-reviewed studies published between 2021 and 2025. The findings reveal a growing interest in applying large language models (LLMs) to high-stakes medical domains, with clinical and educational applications accounting for 60% and 40% of the studies, respectively. Clinically, ChatGPT has shown utility in structured diagnostic tasks such as tumour classification and triage decision support, achieving accuracy rates between 50% and 70%. However, its complex clinical reasoning and multimodal data integration limitations restrict its independent deployment. In education, ChatGPT demonstrated moderate success on licensing exams, with accuracy up to 67%, and was positively received by trainees for foundational learning support. Ethical and epistemological concerns—including AI hallucinations, lack of transparency, bias, and overreliance—were reported in over 70% of the studies. Only 17% of studies met low-risk-of-bias criteria, emphasising the need for stronger methodological rigour. The review concludes that ChatGPT holds transformative potential as a supplementary tool but requires longitudinal validation, regulatory guidance, and interdisciplinary oversight to ensure safe and practical integration. As neurosurgery evolves in the AI era, responsible deployment of ChatGPT must prioritise clinical reliability, educational integrity, and ethical accountability.