Artificial Intelligence, Governance and the Public Interest in Digital Societies

Abstract

Artificial intelligence is becoming embedded in institutional decision-making, public services, education, health systems, finance, regulation, and digital platforms. Its public significance therefore depends not only on model capability but on the organizational, legal, and social arrangements through which automated systems are selected, configured, supervised, contested, and evaluated. This editorial argues that governing AI in the public interest is fundamentally a problem of institutional capacity and accountability. Bias, opacity, data quality, automation bias, privacy risks, vendor dependence, weak procurement, and fragmented responsibility can undermine otherwise capable systems, while formal transparency alone may fail when affected people cannot understand or challenge decisions. Effective governance requires independent auditing, meaningful documentation, contestability, human oversight, public-sector expertise, procurement standards, longitudinal evaluation, and evidence about actual social outcomes after deployment. Identical technologies can produce different consequences across organizations because professional norms, legal safeguards, incentives, data infrastructures, and administrative practices shape how AI is used. Research should therefore evaluate AI as a socio-technical institutional system rather than as an isolated technical artefact.