MONOGRAPHY: Adopting Without Governing: Artificial Intelligence in Slovak Enterprises

Artificial intelligence moved from novelty to normal in the world’s enterprises between 2023 and 2026, yet the organizational practices that should accompany responsible adoption — governing the technology and measuring its value — have lagged conspicuously behind. This monograph examines that lag in a Central-European, small-and-medium-enterprise (SME) setting that the dominant Western, large-firm literature has largely overlooked. Drawing on the AI-ImpactSK project, it analyzes 58 semi-structured interviews with Slovak organizations collected in two waves (autumn 2025 and spring 2026) and funded by the European Union’s NextGenerationEU Recovery and Resilience Plan. The analysis combines framework-based reflexive thematic analysis (eight categories, twenty-four themes, 629 coded segments) with a computational triangulation of 58 English-translated transcripts and a suite of association tests with effect sizes, bootstrap confidence intervals, and multiplicity control.

Three findings define the adoption regime. First, a governance gap: explicit AI governance appears in only 13.8% of interviews and regulation in 6.9%, with governance constituting just 1.9% of all coded content; the gap is systemic rather than confined to any firm subgroup, and governance surfaces as diffuse, informal vigilance rather than codified practice. Second, a measurement paradox: 65.5% of firms articulate concrete benefits from AI, but only 20.7% measure them, and just 10.5% of benefit-claimers measure their gains (McNemar p < .001) — adoption proceeds largely without evaluation. Third, configurations of practice: adoption is informal and bottom-up by default, information-and-communication-technology firms show modestly higher implementation formality, and four indicative adopter configurations emerge from a coding-independent typology. A methodological contribution accompanies these results: the study demonstrates AI-assisted framework coding triangulated with computational text analysis, and transparently reports a wave-level coding-depth artifact while isolating a robust, coding-independent signal — the sharp rise of agentic-AI discourse (mentioned by 37.9% of firms in 2025 and 69.0% in 2026). The monograph argues that Slovak — and likely wider CEE/SME — firms are adopting without governing or measuring: a legitimacy-driven, capability-thin pattern that leaves value undefended and risk unmanaged on the eve of the EU AI Act.

Authors:

Peter Štetka, Nora Grisáková