Reported and anticipated workforce reconfiguration during artificial intelligence adoption: Firm-level evidence from Slovakia
TL;DR
This study examines how AI adoption stage, self-assessed AI maturity, orientation, and ownership are associated with the incidence and co-occurrence of realized or planned job creation and elimination of AI-engaged Slovak firms.
Abstract
Type of the article: Research ArticleAbstractArtificial intelligence (AI) is reshaping how firms design work, yet firm-level evidence compresses employment implications into a single net figure that hides simultaneous hiring and cutting within one firm. The study examines how AI adoption stage, self-assessed AI maturity, orientation, and ownership are associated with the incidence and co-occurrence of realized or planned job creation and elimination. The analytical sample is 693 AI-engaged Slovak firms from a cross-sectional survey (351 current users, 211 pilot firms, 131 planning adoption). Two binary items record whether AI-related positions have been introduced or are planned, and whether AI has led or is expected to lead to layoffs. Each item merges realized with planned or expected action, so the outcomes indicate the incidence of an actual or anticipated event, not the number of positions. We applied exact tests, logistic and multinomial logistic regression, and a bivariate probit model with Benjamin-Hochberg false-discovery-rate (FDR) adjustment. Reported or planned creation was more frequent than elimination (18.0% vs. 9.2%; p < 0.001), and the two co-occurred well beyond chance (odds ratio 7.23; ρ = 0.54). A more advanced adoption stage was associated with a higher incidence of creation (OR 1.49; FDR-adjusted p = 0.004), and employee job-threat concern was associated with a higher incidence of elimination (OR 1.58; p = 0.014). Foreign ownership was associated with joint occurrence of both outcomes in an exploratory model only (RRR 2.29; unadjusted p = 0.027). The study contributes a descriptive typology of four incidence patterns; the cross-sectional design supports associational reading only.AcknowledgmentsThe authors thank the AI-ImpactSK project team and all responding firms.This paper was supported by the European Union – NextGenerationEU through the Recovery and Resilience Plan for Slovakia under project No. 09I05-03-V02-00003/2025/VA (AI-impactSK); 100% share.