数据显示，30%至55%的成年人曾利用ChatGPT或其他大语言模型处理健康问题 1。英国的一项调查显示，超过三分之一的受访者使用聊天机器人来决定是否就诊全科医生（GP） 1。此外，2026年针对Microsoft Copilot超过50万次健康对话的分析显示，约五分之一的互动内容涉及个人症状及治疗方案 1。然而，这种便利性背后潜藏着显著的安全隐患，一项涵盖222个初级保健问题的研究显示，5%至13%的人工智能回复存在可能导致严重伤害的不安全性 1。
此类风险已造成实际悲剧，一名31岁的女性Bowen Wang于2月因听从ChatGPT提供的错误癫痫药物减量建议而死亡 1。尽管后果可能严重，但医患之间的透明度极低，波兰的一项调查显示高达83%的患者未向医生披露其使用AI聊天机器人的行为 1。为应对这一挑战，专家提出遵循“SMART”五步指南以确保咨询安全：即Slow down（慢下来）、Make questions clear（问题清晰）、Ask yourself what am I sharing（思考分享什么）、Remember to check the facts（核实事实）以及Talk to a real person（咨询真人） 1。相关机构呼吁加强医患沟通机制，并对AI在医疗决策中的使用实施更严格的监管指导 1。
A growing number of adults are turning to generative artificial intelligence, such as ChatGPT, for health-related inquiries. Between 30% and 55% of adults have utilized these large language models to address health questions 1. The reliance on AI extends to decision-making regarding professional care; a survey in the United Kingdom revealed that more than one-third of respondents used chatbots to determine whether they needed to see a general practitioner 1. Additionally, an analysis of over 500,000 health conversations conducted via Microsoft Copilot in 2026 found that one-fifth involved discussions about personal symptoms and treatments 1.
Despite the convenience offered by these tools, experts warn of significant safety risks associated with inaccurate or misleading information. A study focusing on 222 primary care queries demonstrated that between 5% and 13% of AI-generated responses contained unsafe content capable of causing serious harm 1. The dangers are not merely theoretical; in February, a 31-year-old woman named Bowen Wang died after following incorrect advice from ChatGPT regarding the reduction of her epilepsy medication dosage 1. Compounding these risks is a widespread lack of transparency, as a survey in Poland indicated that 83% of patients did not disclose their use of AI chatbots to their doctors 1. To mitigate these issues, specialists recommend adhering to the "SMART" guidelines, which advise users to Slow down, Make questions clear, Ask yourself what am I sharing, Remember to check the facts, and Talk to a real person 1.
评论
还没有评论，欢迎留下第一条。