Principles alone cannot guarantee ethical AI

Principles alone are insufficient: While 84+ public-private initiatives have produced AI ethics principles that appear to show global consensus, the paper warns that high-level principles cannot guarantee ethical AI without additional mechanisms—a lesson that differs from medicine, where principles have proven more effective.

AI lacks medicine's foundational structures: Unlike medical ethics, AI development is missing four critical elements that make principles work in medicine: (1) shared professional goals and duties to patients, (2) centuries of established professional norms, (3) proven methods to translate principles into actual practice, and (4) strong legal and professional accountability systems.

Consensus may mask deeper disagreements: The apparent agreement on AI ethics principles across organizations may create a false sense of progress by hiding fundamental political and normative disagreements about what ethical AI actually means in practice.

Translation gap is a major problem: The paper identifies that AI ethics has no proven methods to convert high-level principles (like "fairness" or "transparency") into concrete actions—a challenge that is more severe in AI than in medicine, where clinical practice has established pathways for principle implementation.

Accountability mechanisms are weak: Current AI governance lacks the robust professional licensing, oversight bodies, and legal enforcement structures that give medical ethics principles real teeth, meaning violations of AI ethics principles often have no meaningful consequences.