Between Knowing and Doing: A Qualitative Study of How Generative AI Reshapes Strategic Decision-Making in Technology-Oriented SMEs

Sammanfattning

Despite the increasing capabilities of generative AI (GenAI), existing research in strategic decision-making (SDM) has predominantly treated GenAI as a decision-support tool focused on cognitive enhancement and decision outcomes, leaving the everyday activities, routines, social dynamics, and roles through which strategic decisions are developed underexplored. This gap is particularly pronounced in small and medium-sized enterprises (SMEs), where SDM is informal and concentrated among few individuals. Drawing on Strategy-as-Practice (SAP) theory, this study examines how GenAI reshapes SDM in practice in technology-oriented Swedish SMEs. Nineteen semi-structured interviews were conducted with senior decision-makers and analyzed through reflexive thematic analysis. The findings reveal three reshaping dimensions. First, a validation paradox, structurally biased toward confirmation, emerges as decision-makers and GenAI simultaneously validate one another, blurring tool affordance and actor agency. Second, dyadic interactive strategizing emerges as decision-makers engage with GenAI iteratively as a human-like counterpart, partially displacing peer consultation during early-phase strategic work. Third, actors contributing to SDM expand organically while formal decision authority remains unchanged, also destabilizing the credibility signals through which strategic contributions were previously evaluated. Taken together, GenAI reshapes SDM by destabilizing conditions through which SDM is performed, evaluated, and legitimized. Theoretically, the study extends the SAP tradition by introducing the validation paradox, refining Jarzabkowski's typology of strategizing modes, and distinguishing between contributing and authorizing practitioners within Whittington's practitioner concept. It also raises practical implications for SMEs as GenAI is increasingly integrated into SDM, including risk of confirmation bias, difficulty of evaluating strategic input, and potential strategic homogenization across firms.