Ethics and Regulation of AI Systems in Medicine

The integration of artificial intelligence (AI) systems into medical practice, specifically in cancer detection, presents unknown opportunities for better diagnoses and treatments for patients.However, with the integration of AI systems into a traditional relationship between healthcare professionals and patients, questions regarding accountability in this expanded relationship arise since traditional standards of medical law and medical ethics are addressed towards a healthcare professional.Against this backdrop, we investigate the necessary capacities to hold each involved party accountable (i.e., the healthcare professional, the patient, the developer, a regulatory oversight, and perhaps a clinical AI expert to support the healthcare professional).For this, we first explore the ethical and regulatory implications of employing AI systems in healthcare.We stress that the possibility of maintaining accountability is of central importance for the acceptability of the implementation of AI systems.As AI systems are often inscrutable and do not allow any party to explain and justify the behavior of the AI system, we examine whether and how explainable AI (XAI) methods can support each party with their accountability obligations.With our considerations, we propose a theoretical model for distributing accountability among each involved party and finally highlight the need for regulatory frameworks that can enable an ethically acceptable development and use of AI systems.