Abstract: Trust plays a crucial role in clinical medical practice. However, artificial intelligence (AI) used 

for medical diagnosis faces the challenge of establishing trust in clinical applications due to the lack of sufficient 

transparency in its algorithms. This challenge can be further divided into two aspects: the “doctor-machine” trust 

and the “patient-machine” trust. Compared to approaches such as “breaking the black box” and viewing AI as 

a “second opinion”, the “human-machine coupling” approach better characterizes the relationship between AI 

and humans. According to this approach, an ideal AI for medical diagnosis should become part of the doctor’s 

cognitive process, forming a tightly coupled cognitive system with the doctor. This makes the system, rather than 

the machine itself, the entity that needs to be trusted. This approach provides a possible solution to the problem of 

trust in medical AI.

Key Words: Medical artificial intelligence; Black box algorithms; Human-machine trust; Human-machine 

coupling

See the full text at: https://jdn.ucas.ac.cn/public/uploads/files/68832e91c2836.pdf