In the USA, a pilot project is underway to use ChatGPT to generate messages to patients on behalf of doctors, which raises serious ethical concerns. Although the messages themselves are checked by medical staff before sending, the very involvement of artificial intelligence in the field of confidential medical data requires extreme caution.

The initiators of the project were major technology companies Epic and Microsoft, and three clinics across America were selected as pilot sites. The essence of the experiment is that ChatGPT gets access to patients' medical records and generates drafts of response messages based on patient requests. Then the medical staff checks the generated text, makes the necessary edits and only after that sends a message on their behalf.

The participants of the pilot note that drafts often require significant refinement, especially in terms of adding empathy and personal treatment, as if the message was made by a real doctor. At the same time, studies have shown the high effectiveness of ChatGPT in answering medical questions on open forums.

The main ethical problem is that patients are not informed about the use of AI to generate messages. This lack of transparency undermines the trust between the doctor and the patient. Experts urge to inform patients and consider such experiments as a special form of social research.

In addition, there are risks of AI errors and the spread of medical misinformation by it. There have already been cases when Epic's algorithms incorrectly predicted the development of diseases. Automatically generated messages can go to patients in unedited form due to the negligence of the staff.

Thus, the introduction of AI into medicine, despite the potential benefits, raises serious ethical questions. Transparency and constant communication with patients are critically important in order to ensure the maintenance of trust and to derive real benefits from the use of artificial intelligence technologies in healthcare.