The rise of artificial intelligence (AI) has brought a surge of ethical concerns, prompting numerous public and private organizations to publish high-level ethical guidelines. However, there remains a pressing need to explore practical ways to help AI developers anticipate, recognize, and address ethical challenges in AI systems. This is particularly critical in healthcare, where AI applications frequently interact directly with patients who may be in vulnerable states. This paper proposes an ‘embedded ethics’ approach, in which ethicists collaborate closely with developers throughout an iterative and continuous development process, as a potential method for integrating robust ethical considerations into the practical design and deployment of medical AI.
