The Missing Dimension in Clinical AI: Making Hidden Values Visible

AbstractArtificial intelligence (AI) tools increasingly support clinical decision-making, yet they embed hidden value frameworks that shape recommendations in ways neither physicians nor patients can anticipate. These systems may prioritize different, sometimes conflicting, objectives: maximizing revenue versus minimizing costs, respecting patient autonomy versus preventing harm, or favoring aggressive intervention versus conservative management. This consensus statement from the Responsible AI for Social and Ethical Healthcare (RAISE) symposium identifies a critical gap in current AI governance: Although model cards and regulatory guidelines describe technical specifications and broad principles, they fail to disclose the clinical values embedded in actual recommendations. Drawing on clinical scenarios where both human experts and large language models demonstrate divergent value-based decisions, we propose a "Values In the Model" (VIM) framework. The VIM is a transparent labeling system that documents how AI systems navigate value-laden clinical trade-offs. Rather than relying on electronic health record data that may perpetuate institutional biases, we recommend developing rigorous benchmarks of clinical scenarios tested against diverse stakeholder perspectives. The VIM would enable health care systems, regulators, and patients to make informed choices about AI alignment with their priorities, transforming hidden commitments into explicit, accountable ones. No existing organization currently possesses the combination of technical expertise, clinical knowledge, ethical standing, and stakeholder representation needed to steward such an effort comprehensively. We therefore call for parallel tracks: public debate on values in medical AI and carefully monitored pilot projects in leading health care systems. Without transparency into embedded values, AI risks entrenching the priorities of developers, payers, or providers at scale. With it, we can build systems that reflect the diverse values that make medicine not only a science but a fundamentally human practice.