AI Governance and Workforce Displacement: Who Bears the Cost?

Sammanfattning

AI adoption is spreading rapidly, and firms respond with voluntary governance frameworks pledging responsible use. Yet companies publishing these commitments simultaneously execute workforce reductions attributed to AI. Whether such governance produces genuine accountability, or primarily legitimises the firms that publish it, is the question this inquiry addresses. Sweden provides the empirical setting, a most-likely case for substantive governance given strong institutional commitments to social sustainability and worker protection. This inquiry takes a pragmatist position, choosing methods for the analytical work they do rather than for fit with a prior doctrine. The study employs a two-phase qualitative design. Phase 1 applies Critical Discourse Analysis to 76 governance documents across twelve AI-adopting firms operating in Sweden. Phase 2 applies Reflexive Thematic Analysis to six practitioner interviews (four HR, two Union). Three pillars structure the analytical framework (institutional isomorphism, legitimacy theory, and externality theory), with Integrative Social Contracts Theory providing the normative anchor. The two phases converge strongly. Corporate AI governance frameworks in Sweden function primarily as legitimacy mechanisms, not as accountability mechanisms for workforce displacement. The gap is reproduced across both documents and practitioner accounts, and across convergent and divergent firm cases. Documents omit labour from governance scope and place AI-investment alongside workforce-reduction language without connecting the two. Firm-captured benefits are named while worker- and public-borne costs are silenced. Practitioners experience governance as a practical instrument for the firm, and inhabit decoupling in their own work. The pattern is structurally produced: competitive markets reward symbolic compliance over the unilateral internalisation of displacement costs. Where institutional conditions most favour substantive governance, the gap still holds. The findings imply that closing the gap is a collective-institutional problem rather than a corporate-virtue problem.