Teaching AI Ethics in Medical Education: A Scoping Review of Current Literature and Practices

The teaching of AI should receive dedicated time in the medical curriculum with a longitudinal approach, including preclinical and clinical education

Ethics of AI (incl. Bias) are proposed to be taught in the clinical years

Systematic teaching of AI in medical education recommended to prepare future medical practitioners sufficiently

Ethical principles regarding the use of AI in medicine and the associated data collection, storage and analysis should be taught within medical education.

Due to the advancements of AI in medicine, AI should be implemented into the medical curriculum

Interdisciplinary research is needed on how to implement AI into medical education

To effectively address the broad ethical challenges introduced by AI in healthcare, instructors should possess a strong competency in bioethics

The current education on AI in medical education is limited

The participating medical students perceived AI as an important topic for their medical education

AI education should facilitate an understanding of AI ethics

An understanding of the ethical challenges related to the use of AI in medicine is crucial to prepare medical students for upcoming challenges of medical practice

The teaching of AI ethics should be based on ethical challenges such as informed consent, bias, safety, transparency, patient privacy, and allocation

Real-life examples and case studies should be used to teach AI ethics

AI ethics teaching should be based on associated ethical issues (e.g., bias)

Teaching should align with existing medical ethics lessons

Technical knowledge be taught as part of ethical lessons, while educating both academic staff and medical students

Although the teaching of AI within medical education is recommended, there are only few implementations reported

Research is needed on how to best implement AI into medical curricula

Medical students and future physicians should receive educations on the emerging ethical challenges related to the use of AI in medicine

Medical education should prepare learners for the potential changes of the use of AI in medicine

Lack of consensus on teaching modalities and content related to AI identified

Curricular content on AI and AI ethics recommended

Curricular and extracurricular learning opportunities on AI technologies should include ethical implications of AI

AI education should be available after medical school

Medical education should help medical students to respond to the ethical challenges that arise due to the use of AI in medicine

Empathy and compassion should be fundamental to the curricular development and teaching related to AI

The topic of AI and related content should be part of the medical curriculum, with a staged approach throughout medical education

Preclinical education should include ethics and legal issues with AI

Machine learning (ML) and data science should be part of medical education

Risks, benefits and ethical issues related to the use of ML in medicine should be taught.

General ethical problems and challenges (not specified) [43, 46, 48, 50]

Informed consent, bias, safety, transparency, patient privacy, and allocation [47, 49]

Quality assurance, trust, patient values, confidentially, justice, human rights, accountability, over-diagnosis and over-treatment, automation bias, skill erosion, explainability, and information overload [49]

Incorrect, absent, or abusive use of AI in medicine [49]

Patient-physician relationship [22, 42, 45]

Cyber-security risks [49]

Risks for patient data due to the use of AI [22, 44, 45, 47]

AI-related data collection, storage, and analysis [51]

Liability in case of mistakes due to programming or construction flaws, lack of proper documentation, and user guidance [49]

Ethical implications of liability in the clinical context (not specified) [42, 44, 45, 46]

The potential impact of AI on the principles of medical ethics by Beauchamp and Childress (beneficence, justice, autonomy, and non-maleficence) [22, 36, 42, 44, 45, 49]

Fairness, transparency, and responsibility analogous to beneficence, justice, autonomy, and non-maleficence [22]

Empathy as the cornerstone of teaching AI ethics [23]