Reconfiguring Expertise: AI, Relational Practice, and Clinical Learning in the Emergency Department

Why It Matters

Understanding AI’s relational impact on clinical expertise guides policy, training, and safety strategies, ensuring technology augments rather than undermines professional judgment.

Key Takeaways

• AI reshapes expertise distribution in Danish emergency departments.
• Ethnography shows clinicians retain judgment despite algorithmic recommendations.
• Different AI types (imaging vs LLM) follow distinct implementation paths.
• Expertise is relational, not a static, automatable commodity.
• Junior specialist roles shrink while senior expertise demand intensifies.

Summary

The Oxford Digital Ethnography seminar featured Professor Maya Brun’s ethnographic investigation of how artificial‑intelligence tools are reconfiguring professional expertise in Danish emergency departments. By focusing on a fracture‑detection algorithm and a sepsis‑risk scoring system, Brun illustrates the varied ways AI enters clinical practice, from tightly regulated imaging tools to broader decision‑support platforms.

Her fieldwork reveals that clinicians continue to exercise primary judgment, with liability and final decisions remaining with doctors despite algorithmic suggestions. The study highlights a paradox: while AI promises to automate routine tasks, it simultaneously reduces demand for junior specialists yet amplifies the need for senior experts who can interpret, validate, and troubleshoot algorithmic outputs.

Brun draws on anthropological concepts of relational expertise, citing Dreyfus’s skill‑acquisition model and the actor‑network perspective to argue that expertise emerges through everyday interactions rather than residing as a static, digitizable asset. Concrete examples—radiograph markup, early‑warning scores, and generative‑AI tools in education—show how different AI forms follow distinct implementation pathways and cultural negotiations.

The findings suggest hospitals must redesign training, governance, and inter‑professional workflows to accommodate AI as a collaborative partner rather than a replacement. Recognizing expertise as relational can mitigate the illusion of fully automated decision‑making and ensure patient safety while leveraging AI’s efficiency gains.

Original Description

An Oxford Digital Ethnography Group Seminar.

"Reconfiguring Expertise: AI, Relational Practice, and Clinical Learning in the Emergency Department" with Prof Maja Hojer Bruun.

Public debates often cast AI as either a threat to professional expertise, risking deskilling and redundancy, or as a technical fix that enhances efficiency and liberates time. In this talk, I challenge such dichotomies by arguing that expertise is not something people have, like a resource to draw on for the successful execution of technical operations that can be automated but learned practices that are continuously evolving through collaborative work in professional communities and through negotiations of power and agency in institutions. Drawing on anthropological and STS debates on knowledge and expertise, as well as anthropological and psychological research on professional learning, I conceptualize expertise as knowledgeable practice: material, relational, and historically situated. The talk draws on ethnographic research in two Danish emergency departments during the rollout of an AI tool for fracture detection. Through interviews and participant observation, I explore how AI reconfigures diagnostic work, supervision, and collaboration. I show how clinicians navigate tensions between efficiency and learning, and how expertise, organizational rhythms and temporal logics are renegotiated across professional boundaries as new actors, including data scientists, managers and procurement staff, enter the clinical scene.

Maja Hojer Bruun is a Professor in the Department of Educational Anthropology, Danish School of Education, Aarhus University where she convenes the research programme Future Technology, Culture and Learning Processes. She edited the Palgrave Handbook of the Anthropology of Technology (2022) and has published extensively on emerging digital technologies based on ethnographic research. Her current research focuses on interprofessional collaborations and emerging forms of expertise involved in the development and use of algorithmic systems and AI.