Despite significant advances in AI and its broad penetration across industries, AI in health care has lagged behind. AI tools are not part of routine patient care, even in technologically advanced countries. This retardation is not due to the lack of technical capacity. In fact, the research innovation in this field with respect to AI continues at rapid speed, encompassing nearly all of the areas of clinical care and population health. Unfortunately, very little of this advancement benefits patients. The gap between what this technology can do and what it actually does in hospitals world-wide continues to widen. What causes this disconnect? One of the key contributing factors is the lack of trust in the AI algorithms by both physicians and patients. Partially fueled by the media’s depiction and by lay public unfamiliarity with machine learning, the known deficiencies of AI technology in the context of life-critical care cause significant concerns across the board. It is unrealistic to expect that AI can be errorfree. However, we hypothesize that by making AI algorithms trustworthy for both professional and lay audiences, we can facilitate their practical adoption in healthcare systems—bringing about urgently needed transformations in the global health care system, reducing costs, and saving lives. The work described here will leverage the unique expertise of MIT and J-Clinic researchers in AI algorithm development, and build on the significant experience of the team in large scale clinical deployment of their tools in hospital systems.