The development of artificial intelligence (AI) systems to support diagnostic decision-making is rapidly expanding in health care. However, important challenges remain in executing algorithmic systems at the frontlines of clinical practice. Hence, most often, these systems have not been trained with local data nor do they fit with context-specific patterns of care. This research examines the implementation of an AI-based decision support system (DSS) in the emergency department of a large Academic Health Center (AHC) in Canada, focusing specifically on the question of end-user adoption. Based in an interpretative perspective, the study analyzes the perceptions of healthcare managers, AI developers, physicians and nurses on the DSS, so as to make sense of the main barriers to its adoption by emergency physicians. The study points to the importance of considering interconnections between technical, human and organizational factors to better grasp the unique challenges raised by AI systems in health care. It further emphasizes the need to investigate actors’ perceptions of AI in order to develop strategies to adequately test and adapt AI systems, and ensure that they meet the needs of health professionals and patients. This research is particularly relevant at a time when considerable investments are being made to develop and deploy AI-based systems in health care. Empirically probing the conditions under which AI-based systems can effectively be integrated into processes and workflow is essential for maximizing the benefits these investments can bring to the organization and delivery of care.