基于“医患信任”理论的医疗AI可信度问题的探讨

On Credibility of Medical Artificial Intelligence Technologies based on Physician-Patient Trust

21世纪以来,人工智能技术在医疗领域中得到了广泛的应用,其在疾病预测方面显示出超出人类医生的精准和高效。然而,由于尚不清楚人工智能系统内部的工作原理,这种精准预测是如何实现的仍然无法解释。这被称为人工智能的“黑箱问题”。“黑箱问题”引发了人们对于人工智能“不透明性”的担忧,进而对其可信度提出了质疑。面对质疑,一种观点认为,对于“黑箱”的担忧并无必要。这种观点的持有者往往强调,很多时候,人类医生是凭着直觉和经验在为病人做诊断,而这些直觉和经验连他们自己也很难解释清楚。既然我们能够信任人类医生,也应该信任人工智能。基于信任理论,从医患信任的角度来回应上述观点,尝试论证,医患信任和患者对于医疗人工智能的信任,这两种信任的基础完全不同,不具有可比性。

Since the 21st century,artificial intelligence(AI)technology has been widely used in the medical field,and has shown more accuracy and efficiency than human doctors for diagnosis.However,as the internal mechanisms are often unclear,an AI system is usually regarded as a“black box.”This“black box”problem has raised concerns regarding the credibility of AI.However,one view claims that the concern regarding the“black box”is unnecessary,as human doctors do not seem not much better than AI in terms of providing diagnoses based on intuition and experience.In many cases,the human intuitions and/or experiences are also difficult to explain.Thus,if we can trust human doctors,it seems reasonable to trust medical AI.In this study,I attempt to respond this view,based on a theory of physician-patient trust.