Invisible gains: how resistance contests the promised social benefits of generative AI in SMEs

Abstract

This study aims to investigate how resistance to generative artificial intelligence (GenAI) emerges in small- and medium-sized enterprises (SMEs) when adoption promises to improve working conditions, and what this reveals about the benefits realised by those who do the work. The study draws on 108 semi-structured interviews across 75 Dutch SMEs, involving founders, managers and operational staff. Sampling was stratified across these roles to triangulate perspectives and reveal where accounts converge, diverge and contest. The analysis coded how actors interpreted GenAI in everyday work. Resistance to GenAI does not stem from technological incapacity but from an evaluability paradox: the technology feels accessible and functional while the organisational distribution of value, accountability and recognition remains opaque. Employees experience this as unrecognised labour, managers as eroded process visibility and owners as misunderstood benefit. Resistance takes the form of calibrated engagement, selective reliance and identity-preserving practices; where organisations redistributed information, preserved relational deliberation or drew boundaries around human-led work, it reconfigured into conditional, productive engagement. The analysis captures GenAI adoption in Dutch SMEs during a formative phase. Comparative and longitudinal research would extend the framework. The study offers SME managers a way to read employee resistance to GenAI as diagnostic feedback rather than obstruction. It provides a role-based checklist linking the signals owners, managers and employees display to their underlying causes and to concrete responses: redistributing system information, making value verifiable at the operational level, establishing protocols for interpreting outputs collectively, formalising feedback loops and drawing explicit boundaries around human-led work. Rather than overcoming resistance, managers are advised to interpret it, since it reveals where accountability, recognition or value distribution have broken down. Acting on these signals supports more durable and accountable GenAI adoption. The study shows that GenAI’s social impact in SMEs is not delivered by the technology but negotiated in everyday work, through how its benefits, burdens and recognition are distributed among the people who do it. Promised improvements to working conditions often remain invisible or unevenly shared, with efficiency claimed organisationally while unrecognised labour accumulates locally. Resistance is how this contested distribution becomes visible. Interpreting it as feedback allows firms to align adoption with decent work (SDG-8) and accountable innovation (SDG-9). The social value of GenAI thus depends less on the tool than on whether organisations respond to those who use it. This study advances understanding of the social impact of GenAI in small business: its benefits, burdens and recognition are not delivered by the technology but contested in everyday work. Resistance functions as situated ethical negotiation rather than opposition, and thus as a critical organisational capacity that, read as feedback rather than failure, enables more accountable, socially beneficial adoption.