The Human Instructor Is Gone - What If We Cast GenAI for the Role?

As the technical ease of deploying Generative AI (GenAI) has led to its rapid adoption in healthcare training workflows, a critical gap has emerged in understanding how it reconfigures key aspects of work practice; e.g., instructional authority, responsibility, and interpretation in CPR training. In this speculative study, we enact three probes in a simulation-based training setup that positions GenAI as: 1) exploratory (e.g., noise); 2) instructional (e.g., context); and 3) procedural (e.g., performance measures). Inspired by Blythe (2023), we present Casting GenAI, which treats AI as a design material by casting it as role-based proxies. Across these probes, we examine how GenAI mediates data, surfacing both opportunities and challenges. We argue that moments of GenAI overreach and breakdown should be understood as productive design signals that expose assumptions embedded in proxy roles. The paper contributes a role-based probe approach for early-stage design inquiry and offers insights for design teams exploring GenAI integration in high-stakes sociotechnical systems.