Artificial Intelligence (AI) is increasingly introduced into healthcare settings, yet its integration into fast-paced, high-pressure domains such as Emergency Medical Services (EMS) remains limited. EMS work unfolds across distinct stages, each characterized by different information needs, constraints, and forms of collaboration. Designing effective AI support requires understanding how AI interventions align with, or disrupt, EMS work across its different stages. We conducted semi-structured interviews with 25 EMS clinicians across the United States to examine how existing technologies currently support emergency services workflows and how they envision opportunities for, and concerns about, future AI-based support across different stages of emergency response. Our analysis reveals the cognitive, social, and procedural factors that enable EMS team coordination, which is grounded in situational awareness across distributed roles. EMS clinicians expressed significant concerns about how AI integration threatens this coordination mechanism across multiple dimensions: legal and privacy issues, technical reliability, contextual sensitivity, professional autonomy, and workflow friction. We propose five design principles for AI systems that augment distributed cognition and situational awareness, enabling EMS teams to deliver effective care under extreme constraints.

翻译：人工智能正逐步渗透至医疗保健领域，但其在急诊医疗服务（EMS）等快节奏、高压环境中的融合依然有限。EMS工作贯穿多个不同阶段，每个阶段涉及不同的信息需求、约束条件与协作形式。设计有效的人工智能支持方案，需理解这些干预措施如何在不同阶段与EMS工作相适应或造成干扰。我们对全美25名EMS临床医生进行半结构化访谈，旨在考察现有技术如何支持急救服务流程，以及他们如何设想过未来在应急响应不同阶段基于人工智能的支持机遇与隐忧。分析揭示了实现EMS团队协调的认知、社会及程序性因素——这一协调机制根植于分布式角色间的态势感知能力。临床医生对人工智能整合可能从法律隐私、技术可靠性、情境敏感性、专业自主性及工作流程摩擦等多个维度破坏这种协调机制表示深切忧虑。我们据此提出五项人工智能系统设计原则，通过增强分布式认知与态势感知，使EMS团队能在极端约束下提供有效救治。