Epistemological Gatekeeping in Technology Startup AI Adoption: A Narrative Review of Epistemic Injustice, Concentrated Leadership, and Algorithmic Bias

Abstract

Technology startups are adopting artificial intelligence at an increasing pace, yet smaller firms are more than twice as likely as larger organizations to lack governance roadmaps or dedicated teams for overseeing AI adoption. In startups with 10 to 100 employees, a small number of founders and C-suite leaders hold concentrated authority over AI decisions, functioning as epistemological gatekeepers who determine whose knowledge shapes adoption choices and whose perspectives are excluded. This narrative literature review integrates scholarship across three fields that have developed largely in isolation: Miranda Fricker’s epistemic injustice theory, entrepreneurship research on startup decision-making and governance, and critical AI studies examining algorithmic bias. The review identifies particular organizational mechanisms through which epistemic injustice operates in startup AI governance. These include inferential inertia, temporal shifts in credibility, hermeneutical closure, and the active promotion of dominant epistemological assumptions by AI systems themselves. Results show that epistemological gatekeeping in startups is not a matter of individual bias but rather a structural phenomenon arising from the interaction of concentrated authority, informal governance, speed culture, resource scarcity, and demographic homogeneity. The review concludes that bias mitigation is fundamentally an epistemological problem requiring attention to whose knowledge informs AI decisions, and recommends phenomenological research to investigate how startup leaders experience and enact their gatekeeping role.