Early AI deployment in a Nordic transit agency revealed that institutional control and worker autonomy co-evolve: governance seeks to constrain risks while employees appropriate AI to build capabilities, producing both tensions and synergies that shape organizational resilience.

Main Finding

Early AI adoption in a Nordic public-transportation organization reveals that digital sovereignty is not a static attribute but an ongoing, bidirectional negotiation between organizational governance (control, compliance, resilience) and individual worker autonomy (agency, discretion, skill development). These opposing goals create persistent tensions but also create synergies: AI introduction can simultaneously constrain and enable workers, and—when managed deliberately—serve as a capability-building process that strengthens both individual autonomy and organizational resilience.

Key Points

• Conceptual contribution: introduces "bidirectional dynamics in digital sovereignties" — a framework treating organizational control objectives and individual autonomy aspirations as mutually influencing, paradoxical tensions rather than one-sided trade-offs.
• Dynamics: digital sovereignty evolves across levels (individual ↔ organizational) during early AI deployment; actors continually negotiate governance, access, and skill boundaries.
• Tensions and synergies: early transformation produces conflicts (e.g., monitoring, standardization, reduced discretion) but also positive feedback loops (e.g., AI-enabled upskilling, improved decision support, faster organizational learning).
• Capability-building: AI adoption can function as a mechanism for building worker capabilities (new competencies, situational autonomy) and organizational resilience (greater adaptive capacity, institutional learning).
• Risks: potential loss of worker agency, opaque decision-making, concentration of control, misalignment between policy goals and on-the-ground practices.
• Practical insight: effective governance requires approaches that balance control and autonomy—participatory design, transparent systems, training, adaptive policies—rather than only top-down regulation.

Data & Methods

• Empirical scope: case study of early-stage AI adoption in a Nordic public-transportation organization (public sector context).
• Methods (reported/typical for this design):
  • Qualitative fieldwork centered on a single organizational setting.
  • Semi-structured interviews with employees, frontline workers, and managers to capture perceptions of autonomy, control, and adaptation.
  • Observation of AI introduction into work practices and review of internal documents/policies to trace governance changes.
  • Thematic and interpretive analysis using a paradoxical/institutional lens to derive the bidirectional dynamics concept.
• Temporal perspective: focuses on early adoption phase, emphasizing evolving goals and emergent negotiation processes rather than mature deployment outcomes.

Implications for AI Economics

• Diffusion and adoption: Digital sovereignty dynamics affect the speed and shape of AI diffusion in organizations. Strong top-down control may speed standardized deployment but can suppress adaptive uses; fostering worker autonomy can enable more diverse, productive uses and local innovation.
• Productivity and complementarities: When AI adoption is managed as capability-building, it can enhance human–AI complementarities, raising labor productivity without simply displacing workers. Policy and firm strategy that invest in training increase returns to AI investments.
• Labor market and bargaining power: Negotiated sovereignty changes the distribution of workplace power—greater autonomy and skill accumulation can strengthen worker bargaining positions; conversely, opaque control regimes may concentrate rents with management or vendors.
• Measurement & evaluation: Economic assessments of AI should account for multi-level sovereignty outcomes (organizational resilience, worker autonomy, governance costs) not just short-term productivity gains. New metrics could track capability growth, autonomy-preserving adoption, and adaptive capacity.
• Policy recommendations:
  • Encourage policies that support participatory governance, transparency, and skills development (public procurement criteria, funding for training).
  • Design regulation mindful of trade-offs: excessive centralization of control can stifle local innovation; too little governance risks fragmentation and inequity.
  • Support longitudinal evaluation of early deployments to capture dynamic effects on welfare, labor outcomes, and organizational adaptability.
• Strategic implications for firms and public organizations: Treat AI adoption as a socio-technical, capability-building process. Balance control mechanisms with investments in worker agency to realize resilient, inclusive economic benefits from AI.