MONOGRAPHY: Artificial Intelligence in Production: Evidence from Slovak Enterprises

Artificial intelligence is expected to reshape how goods are made and how operational processes are run, but the firm-level evidence on what “AI in production” actually amounts to (particularly in the manufacturing-intensive economies of Central and Eastern Europe) is still thin. This monograph uses micro-data from the AI-IMPACT survey of enterprises operating in Slovakia (N = 816 organisations, fielded in 2025) to examine the firms that reported having begun implementing or testing AI in “production and operational processes” (item D3 of the questionnaire). There are 157 such firms: 19.2% of the achieved sample and 27.9% of the 562 firms that use or pilot AI in any form. For these 157 firms the survey carries a dedicated battery of seventeen concrete operational AI applications, perceived positive and negative impacts, seven AI-maturity indicators, perceived barriers, corporate

sentiment, workforce effects, and full firmographic detail, which together allow a reasonably detailed, multi-angle description of production AI in one national economy.

The study combines descriptive profiling with hypothesis testing across eight lenses. Four results recur. First, production AI is not confined to the manufacturing sector: only 40.8% of the adopting firms are classified in manufacturing, with construction, ICT and a range of operationally intensive services accounting for the rest. Among the firmographic factors measured here, belonging to a goods-producing sector is the strongest correlate of adoption (46.4% of goods-producing adopters versus 19.0% of service adopters; adjusted odds ratio 3.67). Second, the factors associated with whether a firm begins production AI differ from those associated with how far it scales: sector and firm size relate to the first, while organisational AI maturity relates to the second (Spearman ρ = 0.35 with a multi-application intensity index; the only significant predictor in a multivariate model, which nonetheless leaves most variance unexplained). Third, firms deploy AI across a wide set of applications rather than around one flagship use case: active use is fairly even across all seventeen applications (28–41%), led by machine-learning process optimisation, quality control and energy monitoring, with predictive maintenance, often treated as the canonical case, near the bottom; this pattern holds when the analysis is restricted to goods-producing firms only. Fourth, the workforce signal leans towards job creation rather than displacement: more firms reported creating or planning new roles (19.1%) than reported or planned layoffs (11.5%), and the difference between the discordant firms is statistically significant (McNemar exact p =.036). Perceived benefits exceeded perceived risks (Wilcoxon p <.001, r = 0.40, robust to a goods-producing-only check); the twenty barrier items loaded on a single factor of very high internal consistency (α =.94), led by integration complexity, skills and a notable gap in state support; and the adopters separated into a larger lower- maturity group (“Operational Experimenters”, n = 107) and a smaller high-maturity group (“Mature Integrators”, n = 50) deploying AI roughly twice as broadly. The adoption wave is recent: 61.1% of adopters dated their first AI project to 2024–2025.

The data are cross-sectional, self-reported and drawn from a voluntary, AI-engaged sample, so the findings are read as associations within an achieved sample rather than as causal estimates for the Slovak population; common-method variance, tested with Harman’s single-factor criterion, was not dominant (largest component 26.7%). Within those bounds the monograph offers an empirically grounded account of production AI as an operational-process phenomenon that spreads across sectors along structural lines and deepens along organisational-readiness lines, and it gives Slovak and CEE policymakers firm-level evidence on where the binding constraints (skills, integration capability and public support) appear to lie.

Authors:

Peter Štetka & Andriana Baziuk