Imagine an 80-year-old with several comorbid conditions seeking emergency care due to breathing difficulties. As this person enters into the emergency care, the medical staff collects basic information regarding both personal social-economic background, medical history and current symptoms. Such data is fed to a computer system that will suggest a diagnosis based on the provided information. Should the medical staff tell the patient that they used an AI system in order to form this diagnosis? What if the physician disagrees with the recommendation and decides on a different diagnosis, should the patient have this information? Should the physician inform about possible biases or faults in the training datasets that may constitute risks for the patient? What if the patient does not even know what is ‘AI’ and how it works? These are some of the practical dilemmas and interrogations that Marika Mäkinen, will attempt to tackle in her doctoral dissertation.

This presentation will introduce her doctoral project concerning the use of AI systems as diagnostic tools and medical informed consent. A project that discusses how the current doctrines of medical informed consent fit in with AI-assisted medical diagnoses, focusing on two main questions: 1) how much and what kind of information should the patient have regarding the AI system when such system is used as a diagnostic tool?; and 2) should the patient get different information when AI recommends a diagnosis compared to a situation where such technologies are not used?