FRAMING FOR AI ADOPTION Understanding Change Agents as Boundary Spanners

Sammanfattning

Organizations increasingly recognize the potential of artificial intelligence (AI), yet many struggle to turn this potential into practical use. Since research has shown that challenges of AI adoption are often organizational rather than technical, organizations often rely on external facilitators for support. However, limited research has examined what these actors do in practice. To explore this, the study examines how external change agents facilitate organizational AI adoption. It draws on qualitative data collected at AI Sweden, including observations of seven workshops and post-workshop interviews with change agents and facilitators. The findings show that change agents first build readiness for AI adoption by surfacing participants' concerns, addressing uncertainty, and establishing legitimacy, thereby creating the conditions for engagement before knowledge work begins. They then translate AI knowledge through metaphors and artifacts that make abstract concepts more understandable and connectable to participants' own work contexts. Finally, they facilitate organizational change by equipping participants with tools and shared language that enable continued AI work independently of external support. Combining boundary spanning theory and technological frames, the study shows that frame alignment precedes and enables boundary spanning. The findings further demonstrate that boundary spanning in early AI adoption can be temporary and exit-oriented, aimed at reducing dependence on external actors. This study contributes to research on AI adoption by providing empirical insights into how facilitation is carried out. It further contributes to boundary spanning theory and the technological frames literature by showing that framing is not a complement to knowledge transfer but its precondition.