GENERATIVE AI AND SELF-MEDICATION: RISKS, RELIABILITY, AND PATIENT SAFETY IN THE ERA OF AI-BASED HEALTHCARE

Engineering notes

Engineering notes will be added by the aipentium editorial team.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为大语言模型、生成式AI、ChatGPT相关技术、计算机视觉、深度学习等高价值论文补充中文说明。

Original abstract

Health-professions students are increasingly using artificial intelligence (AI) tools, especially large language model (LLM)-based chatbots like ChatGPT, for academic tasks as well as informal drug information retrieval, symptom checking, and self-medication decision-making.[2,9,3,7] Due to their dual roles as future gatekeepers of safe pharmaceutical use, trainees in sciences connected to medicine, and consumers of health information, pharmacy students hold a special place in this conversation.[2,3,20,33,56] In order to describe what is known about pharmacy (and allied health) students' knowledge of AI, their attitudes toward its use in clinical and self-care contexts, and their actual practices—including the use of AI for drug information, dosing guidance, and self-diagnosis-adjacent tasks—this review synthesizes nine cross-sectional Knowledge, Attitude, and Practice (KAP) studies conducted between 2022 and 2026 across Zambia.[3] Saudi Arabia.[2,6] India.[1,4,8] Syria.[9] Malaysia.[5] and other settings. AI awareness is almost universal (82–100%) in all contexts.[2,3,5,6,8,9] but conceptual understanding of AI subtypes, medical uses, and limitations is generally lower (30–60%).[2,3,5,6,8,9] Concerns concerning data privacy, false information, over-reliance, and professional displacement temper the generally positive attitudes.[1,2,6,7,8] Practice is still limited: formal curricular training is uncommon (7–46%).[2,3,6,7,9] and the majority of self-reported AI use focuses on academic activities (summarizing, presentations, exam preparation) rather than verified clinical decision-support.[3,4,5,7] There is a dearth of direct data on AI-assisted self-diagnosis and self-medication, particularly among pharmacy students. In order to outline the emerging risk landscape and suggest curricular and regulatory responses, this review draws conclusions from related findings, such as the use of ChatGPT for "getting drug information".[2] perceived AI usefulness in "medication management".[1,2] and expressed willingness to trust AI-generated dosing recommendations.[2]

Links and sources

Comments