Artificial Intelligence Adoption and Labour Productivity in Slovakia and the EU27: Implications for Sustainable Economic Growth

Abstract

This study analyses the adoption of artificial intelligence (AI) in enterprises in Slovakia in comparison with the EU27 and examines its relationship with labour productivity from the perspective of long-term economic sustainability. Using harmonised Eurostat data for the period 2021–2024, the analysis applies descriptive statistics, gap analysis, dynamics of change, correlation analysis, and an illustrative regression model. The results show that although AI adoption in Slovakia increased across all enterprise size classes, it consistently remained below the EU27 average. Labour productivity developments in Slovakia were characterised by substantial short-term volatility and did not show a stable association with AI diffusion. Both correlation and illustrative regression results confirm the absence of an immediate statistical relationship between AI adoption and productivity at the aggregate level. These findings suggest that potential productivity improvements associated with AI adoption are likely to depend on complementary investments in organisational transformation, digital skills, and institutional capacity. The study provides empirical evidence for a small open economy within the EU and offers policy-relevant insights into how AI adoption is more likely to support long-term economic sustainability than short-term performance gain.

1. Introduction

The accelerating diffusion of artificial intelligence (AI) technologies across the European economy represents one of the defining transformations of the current industrial era. AI is widely recognised as a general-purpose technology with the potential to enhance productivity, stimulate innovation, and contribute to long-term economic sustainability [1,2,3]. At the same time, its integration into national economies is proceeding unevenly across regions and sectors, leading to widening digital divides both within and between countries of the European Union (EU) [4,5]. For smaller and structurally less diversified economies, such as Slovakia, the capacity to adopt and effectively utilise AI-based solutions will play a critical role in determining their future competitiveness and productivity growth [6].

The European Union has explicitly positioned AI as a cornerstone of its digital and industrial strategies [7]. Through policy frameworks such as the Coordinated Plan on Artificial Intelligence (2021 Review) and the Digital Europe Programme, the EU seeks to accelerate the uptake of AI technologies while ensuring ethical use and inclusiveness across Member States [7,8,9]. These initiatives reflect the strategic expectation that AI adoption can serve as a catalyst for improving labour productivity, efficiency, and resilience, thereby supporting sustainable long-term growth [2]. Yet, despite the growing consensus on the transformative potential of AI, empirical evidence on how adoption patterns relate to productivity outcomes remains limited—especially in smaller EU economies [10].

In the context of Central and Eastern Europe (CEE), several studies and policy reports point to persistent gaps in digital transformation relative to Western Europe [11,12]. These disparities are driven not only by differences in infrastructure and R&D intensity but also by structural and institutional factors, such as firm size distribution, managerial capabilities, and skills endowments [13]. For Slovakia, these challenges are particularly salient: although the country benefits from strong industrial capacity and integration into European value chains, it lags behind the EU27 average in the use of advanced digital technologies, including AI, cloud computing, and big data analytics [14].

From a macroeconomic perspective, labour productivity represents a crucial indicator of economic performance and sustainability [15]. Productivity gains are essential for maintaining competitiveness, supporting wage growth, and ensuring fiscal stability in ageing societies [15,16]. The introduction of AI into business processes—whether through automation, predictive analytics, or process optimisation—is expected to enhance productivity by increasing efficiency and reducing transaction costs [17]. However, in practice, the relationship between AI adoption and productivity growth has proven complex and non-linear. Several authors highlight the existence of a “productivity paradox of AI,” where widespread technological diffusion does not immediately translate into measurable productivity gains [2,18].

The paradox arises partly because productivity improvements depend on complementary factors such as organisational adaptation, workforce reskilling, and the accumulation of intangible assets [19]. Firms adopting AI technologies must restructure workflows, redesign decision processes, and invest in data governance systems before efficiency benefits become visible [20,21]. As a result, the economic impact of AI often emerges with a temporal delay [22]. This lag is particularly evident in less digitally mature economies, where structural barriers—such as limited access to high-quality data, low digital literacy, and scarce capital for innovation—constrain firms’ ability to reap the full benefits of technological adoption [22,23].

Against this backdrop, the present study aims to provide a comparative assessment of AI adoption and labour productivity in Slovakia and the EU27. By analysing harmonised Eurostat datasets, it seeks to describe observable patterns, quantify the magnitude of gaps, and compare their developments over time [24]. The research contributes to the literature in three primary ways. First, it provides empirical evidence for Slovakia, a small, open economy within the EU that remains underrepresented in international studies on AI adoption [25]. Second, it offers a comparative European perspective, situating Slovak developments within broader EU trends. Third, it illustrates potential implications for long-term economic sustainability, highlighting the strategic importance of digital capacity-building as a driver of future productivity growth [26].

Although previous analyses have explored AI readiness and digital intensity across EU Member States [4,27,28], few have focused on the relationship between AI diffusion and productivity in a comparative framework. Moreover, existing studies often rely on survey data or firm-level case studies that provide valuable qualitative insights but limited quantitative comparability [29]. The use of harmonised Eurostat indicators in this paper addresses this limitation by allowing consistent cross-country comparison between Slovakia and the EU27 aggregate [24].

From a policy standpoint, understanding these relationships has critical implications for sustainable growth and convergence [8]. The Slovak economy, characterised by a high degree of industrial specialisation and export dependence, faces the dual challenge of maintaining competitiveness while transitioning towards a more knowledge-intensive growth model [30]. In this context, the diffusion of AI represents both an opportunity and a risk: an opportunity to enhance productivity and efficiency, but also a risk of deepening structural asymmetries if adoption remains concentrated among large firms or specific sectors [31]. Bridging this digital divide requires coordinated efforts across education, innovation policy, and digital infrastructure development.

The analysis also intersects with the broader debate on the economic sustainability of digital transformation [26]. AI adoption can contribute to sustainable growth not only by improving productivity but also by enabling smarter resource allocation, energy efficiency, and circular economy practices [32]. However, sustainability in this sense is conditional on inclusive access to technology, ethical governance, and the capacity to mitigate the social risks associated with automation [2]. For Slovakia and other CEE countries, aligning AI-driven productivity improvements with long-term sustainability objectives—such as social cohesion, employment stability, and technological sovereignty—remains a key policy challenge [33].

The motivation for this study, therefore, lies in bridging empirical and conceptual gaps. While there is a growing policy discourse promoting AI as a source of competitiveness [34], robust quantitative evidence on how AI adoption and productivity evolve in aggregate data remains scarce [18]. By analysing data for the period 2021–2024, this paper provides a descriptive and exploratory perspective on the evolving relationship between digital transformation and economic performance. The analysis does not aim to establish causal inference but rather to identify patterns and potential associations that can inform future research and policy.

The structure of the paper is as follows: After this introduction, the Section 2 describes the data sources, variable construction, and analytical procedures employed, including descriptive statistics, gap and correlation analysis, and an illustrative regression model. The Section 3 presents the empirical findings for Slovakia and the EU27, followed by the Section 4, which situates the results within the broader literature on digital transformation and productivity dynamics. Finally, the Section 5 summarises the main insights and outlines implications for policy and future research.

In summary, this study situates the Slovak experience of AI adoption within the wider European context and contributes to the growing empirical literature on the relationship between digital transformation and productivity [35]. By focusing on cross-national comparison and the quantitative assessment of adoption gaps, it provides evidence-based insights relevant for both policymakers and academic researchers concerned with the economic sustainability of digitalisation [36].

2. Materials and Methods

This study provides a comparative assessment of the adoption of artificial intelligence (AI) technologies in enterprises in Slovakia and the EU27 and examines how these developments relate to selected economic indicators, particularly labour productivity. The purpose of the analysis is to identify and quantify differences between Slovakia and the EU27 in AI adoption and labour productivity and to explore potential associations between these indicators using available data from Eurostat (Luxembourg, Luxembourg). Because AI indicators were collected only in the survey years 2021, 2023, and 2024, the study applied a descriptive and exploratory approach rather than causal inference. No harmonised Eurostat data on AI adoption were available for 2022, as the enterprise survey on the use of digital technologies was not conducted or published for that year. The methodological framework combined descriptive statistics, gap analysis, and comparative trend evaluation to document differences between Slovakia and the EU27. Given the limited number of available observations, the study did not attempt to estimate statistically reliable or inferential relationships or causal effects. The objective was strictly descriptive: to identify observable patterns in AI adoption and labour productivity within the available dataset.

2.1. Research Framework

A workflow diagram (Figure 1) was developed to summarise the analytical procedure applied in the study. The research design follows a sequence of descriptive and comparative steps, progressing from data collection to the evaluation of differences and trends between Slovakia and the EU27. Each step is described in detail below.

The diagram provides a concise visualisation of the methodological structure, showing how the analysis proceeds from initial data processing to the evaluation of differences and descriptive relationships between Slovakia and the EU27.

2.1.1. Step 1: Data Acquisition

Harmonised Eurostat datasets were used to obtain comparable data for Slovakia and the EU27. Three datasets were selected:

• AI adoption in enterprises (isoc_eb_ai, indicator E_AI_TANY), available for 2021, 2023, and 2024. This dataset measures the share of enterprises using at least one AI technology and provides breakdowns by enterprise size (10–49, 50–249, 250+ employees).
• Labour productivity per hour worked (tipsna70), expressed as year-on-year percentage change, available for 2021–2024. This indicator captures short-term developments in economic performance.
• Real GDP growth (tec00115), used as macroeconomic background, available for 2021–2024.

All datasets were retrieved from the Eurostat Data Browser and reflect harmonised EU methodology, enabling direct cross-country comparison [1,2,3].

2.1.2. Step 2: Variable Construction

To prepare the dataset for analysis, all retrieved Eurostat indicators were transformed into analytically consistent variables. AI adoption was expressed as the percentage of enterprises using at least one AI technology, separately for Slovakia and the EU27, and, where available, also by enterprise size classes (10–49, 50–249, and 250+ employees). Labour productivity was taken as the annual percentage change in productivity per hour worked, enabling a consistent measurement of short-term performance.

To quantify Slovakia’s relative position, gap variables were constructed as simple arithmetic differences between Slovak and EU27 values. The AI adoption gap was computed as

$$A{I\_}_{{gap}_t}={AI}_t^{SK}-{AI}_t^{EU}$$

where

• $A{I}_t^{SK}$ = AI adoption in Slovakia in year t;
• $A{I}_t^{EU}$ = AI adoption in the EU27 in year t;
• $A{I\_}_{{gap}_t}$ = difference between Slovakia and the EU27.

The productivity gap was defined analogously:

$$PRO{D}_{{gap}_t}={PROD}_t^{SK}-{PROD}_t^{EU}$$

where

• $PRO{D}_t^{SK}$ = labour productivity growth in Slovakia in year t;
• $PRO{D}_t^{EU}$ = labour productivity growth in the EU27 in year t;
• ${PROD}_{{gap}_t}$ = difference between Slovakia and the EU27.

Inter-year change variables were constructed to capture the dynamics of development:

$$∆{AI}_t={AI}_t-{AI}_{t-1}$$

$$\Delta {PROD}_t={PROD}_t-{PROD}_{t-1}$$

where

• ${AI}_t$ = level of AI adoption in year t;
• ${AI}_{t-1}$ = level of AI adoption in the previous year (t − 1);
• ${∆AI}_t$ = year-to-year change in AI adoption in year t;
• ${PROD}_t$ = labour productivity growth in year t;
• ${PROD}_{t-1}$ = labour productivity growth in the previous year (t − 1);
• ${∆PROD}_t$ = year-to-year change in labour productivity in year t.

For the illustrative regression model, labour productivity served as the dependent variable and AI adoption as the explanatory variable. The model followed the standard linear specification:

$${PROD}_t=\alpha +\beta ·{AI}_t+{\epsilon }_t$$

where

• $\alpha$ = intercept;
• $\beta$ = slope coefficient;
• ${AI}_t$ = level of AI adoption in year t;
• ${\epsilon }_t$ = error term;
• ${PROD}_t$ = labour productivity growth in year t.

2.1.3. Step 3: Descriptive Comparison

The third step consisted of a descriptive comparison of the selected indicators to obtain an initial overview of the differences between Slovakia and the EU27. AI adoption levels were compared across the available years to observe the development of digitalisation in enterprises, both in aggregate and across enterprise size classes. Labour productivity developments were analysed over the 2021–2024 period to identify broader economic trends relevant for interpreting the results.

The descriptive assessment also included a visual inspection of trajectories, allowing for the identification of patterns such as increases, declines, or stability in AI uptake and productivity. In addition, GDP growth was used to contextualise the observed developments by indicating broader macroeconomic conditions in the respective years. This descriptive step serves as the empirical foundation for the subsequent analytical procedures by highlighting baseline differences and visible trends in the data.

2.1.4. Step 4: Gap Analysis

Gap analysis was applied to quantify the absolute differences between Slovakia and the EU27 in both AI adoption and labour productivity. The method relies on the gap variables defined in Step 2, which express the difference between Slovak and EU27 values in each year. By comparing Slovakia directly to the EU27 benchmark, the analysis highlights whether the national level of AI adoption or productivity is above or below the European average.

This step provides a clear comparative perspective and establishes the basis for interpreting subsequent developments and relationships within the dataset.

2.1.5. Step 5: Dynamics of Change

The fifth step focused on analysing the dynamics of change in AI adoption and labour productivity. Year-to-year differences, defined in Step 2, were used to assess the speed and direction of developments in both indicators. This approach makes it possible to identify whether increases or declines in AI adoption coincide with shifts in productivity across the observed period.

By examining the inter-year changes rather than only levels, this step provides additional insight into the momentum of technological adoption and economic performance. The resulting change variables form an intermediate analytical layer that supports the descriptive interpretation of patterns observed in the dataset.

2.1.6. Step 6: Correlation Analysis

The sixth step assessed the degree of association between AI adoption and labour productivity using Pearson’s correlation coefficient. This method provides an indicative measure of whether higher or lower levels of AI adoption tend to coincide with corresponding movements in productivity. Although the limited number of available AI observations restricts the interpretability of the results, the correlation analysis offers an initial insight into the co-movement of the two indicators.

Pearson’s correlation coefficient was calculated using the standard formula:

$$r={\displaystyle \frac{\sum (AIt-\overline{AI})(PRODt-\overline{PROD})}{\sqrt{{\sum (AIt-\overline{AI})}^2}\sqrt{\sum {(PRODt-\overline{PROD})}^2}}}$$

where

• $A{I}_t$ = AI adoption in year t;
• $PRO{D}_t$ = labour productivity in year t;
• $\overline{AI}$, $\overline{PROD}$ = mean values of the respective indicators;
• $r$ = Pearson’s correlation coefficient measuring the strength and direction of the linear association between AI adoption and labour productivity.

The correlation results are interpreted only as descriptive evidence of association and not as causal effects.

2.1.7. Step 7: Illustrative Regression

In the seventh step, a simple linear regression model was used to explore the indicative relationship between AI adoption and labour productivity. As described in Step 2, the model specifies labour productivity as the dependent variable and AI adoption as the explanatory variable. Given the limited number of available observations, the regression is not intended to estimate causal effects or support prediction but merely to describe the mathematical association observed within the limited dataset. The coefficient is reported only to characterise the direction of co-movement in the observed values. The regression results are therefore interpreted only as mathematical outputs and not as evidence of causal impact.

3. Results

The results are structured according to the analytical steps described in Section 2. The findings first present the levels and developments of AI adoption and labour productivity in Slovakia and the EU27, followed by the outcomes of the gap analysis and the assessment of year-to-year changes. The emphasis is placed on descriptive comparison and observable trends within the available data.

3.1. AI Adoption in Slovakia and the EU27

AI adoption in enterprises increased steadily between 2021 and 2024 in both Slovakia and the EU27, although Slovakia consistently remained below the European average (

Table 1. AI adoption in enterprises (%) by size class (Slovakia vs. EU27). Source: Eurostat [1], authors’ processing.

Size Class	Year	Slovakia	EU27
10–49 employees	2021	4.11	6.12
	2023	5.98	6.41
	2024	8.78	11.21
50–249 employees	2021	7.01	12.55
	2023	8.57	13.07
	2024	15.73	20.97
250+ employees	2021	19.44	28.41
	2023	21.89	30.48
	2024	29.10	41.17

). In a broader EU context, Slovakia belongs to the group of countries with below-average levels of AI adoption, while the highest adoption rates are observed in a limited number of digitally advanced EU economies.

Among small enterprises (10–49 employees), adoption in Slovakia rose from 4.11% in 2021 to 8.78% in 2024, compared with an increase from 6.12% to 11.21% in the EU27. A similar pattern was observed among medium-sized enterprises (50–249 employees), where Slovakia grew from 7.01% to 15.73%, while the EU27 increased from 12.55% to 20.97%.

Large enterprises (250+ employees) exhibited the highest levels of adoption. In Slovakia, the rate increased from 19.44% to 29.10%, whereas the EU27 rose from 28.41% to 41.17%. Despite the upward trend in Slovakia, the distance from the EU27 benchmark remained substantial, particularly among large enterprises, indicating slower diffusion of advanced digital technologies. It should be noted that small enterprises (10–49 employees) account for the largest share of the enterprise population in both Slovakia and the EU27, while large enterprises represent only a small fraction. Therefore, AI adoption rates among small firms are particularly important for assessing the overall diffusion of AI technologies in the economy.

3.2. Labour Productivity Developments

Labour productivity growth displayed substantial year-to-year variation in Slovakia, reflecting broader macroeconomic fluctuations (

Table 2. Labour productivity growth (%), Slovakia vs. EU27. Source: Eurostat [2], authors’ processing.

Year	Slovakia	EU27
2021	5.6	0.6
2022	−3.0	0.4
2023	0.9	−0.8
2024	1.7	0.2

). Productivity sharply declined in 2022 (−3.0%), followed by a moderate recovery in 2023 (0.9%) and 2024 (1.7%). In contrast, productivity in the EU27 remained relatively stable with smaller changes, fluctuating between −0.8% and 0.6% over the examined period.

These differences suggest higher volatility in Slovakia’s short-term economic performance, resulting in alternating years of strong and weak productivity dynamics.

3.3. Gap Analysis: Slovakia vs. EU27

The gap analysis highlights systematic underperformance of Slovakia in AI adoption relative to the EU27 across all years and enterprise sizes. For small enterprises, the AI adoption gap ranged from −2.01 p.p. in 2021 to −2.43 p.p. in 2024. The gaps were even wider among medium-sized and large enterprises.

In contrast, productivity gaps showed a mixed pattern (

Table 3. Gaps between Slovakia and EU27 (p.p.). Source: Eurostat [3], authors’ calculations.

AI Gap (Slovakia—EU27)—10–49 Employees
Year	AI Gap
2021	−2.01
2023	−0.43
2024	−2.43
Productivity Gap (Slovakia—EU27)
Year	PROD Gap
2021	+5.0
2022	−3.4
2023	+1.7
2024	+1.5

). Slovakia outperformed the EU27 in 2021 (+5.0 p.p.) and 2023 (+1.7 p.p.), while falling below the EU average in 2022 (−3.4 p.p.). This indicates that productivity does not mirror the AI adoption gap directly, supporting the need for descriptive rather than causal interpretation.

3.4. Dynamics of Change (ΔAI and ΔPROD)

The year-to-year changes demonstrate that Slovakia made meaningful progress in AI adoption, especially between 2023 and 2024 (ΔAI = +2.80 p.p.). However, the EU27 experienced an even larger increase in this period (+4.80 p.p.), suggesting that Slovakia’s relative position did not improve.

Changes in productivity show a different pattern (

Table 4. Year-to-year changes (ΔAI, ΔPROD). Source: Eurostat [1,2], authors’ calculations.

ΔAI (10–49 employees)
Period	ΔAI (SR)	ΔAI (EU27)
2021 → 2023	+1.87	+0.29
2023 → 2024	+2.80	+4.80
ΔPROD
Period	ΔPROD (SR)	ΔPROD(EU27)
2021 → 2022	−8.6	−0.2
2022 → 2023	+3.9	−1.2
2023 → 2024	+0.8	+1.0

). Slovakia experienced a strong decline in 2022, followed by a gradual improvement in the subsequent years. These divergent paths between AI and productivity underscore the importance of analysing the indicators separately and without causal claims.

3.5. Correlation Analysis

The correlation analysis revealed no consistent association between AI adoption and labour productivity in the examined period. Slovakia showed a moderately negative correlation (r = −0.70), while the EU27 displayed a weak positive correlation (r = 0.19) (

Table 5. Pearson correlation between AI adoption and labour productivity. Source: Authors’ calculations.

Country	r
Slovakia	−0.70
EU27	+0.19

).

Given the very small number of data points, these values must be interpreted strictly as descriptive indicators rather than evidence of meaningful statistical relationships.

3.6. Illustrative Regression Results

The illustrative regression model produced a negative slope coefficient for Slovakia (β = −0.746), indicating that productivity tended to fall in years with higher AI adoption levels. For EU27, the coefficient was positive but close to zero (β = 0.048) (

Table 6. Regression coefficients (illustrative model). Source: Authors’ calculations.

Country	α (Intercept)	β (Slope)
Slovakia	7.42	−0.746
EU27	−0.38	+0.048

).

These coefficients reflected only the mechanical relationship within the observed values and were not used for inference, prediction, or scenario construction. Given the extremely small number of observations, standard inferential statistics such as standard errors, confidence intervals, p-values, or goodness-of-fit measures would not be reliable or meaningful. For this reason, they are intentionally not reported. The coefficient is presented purely as a numerical artefact reflecting the mechanical relationship within the limited dataset.

4. Discussion

The results of this study indicate several important patterns related to the adoption of artificial intelligence (AI) in Slovak enterprises and its potential association with labour productivity. The findings confirm that Slovakia continues to lag behind the EU27 in AI adoption across all enterprise size groups, despite gradual progress. This long-term lag in the diffusion of advanced digital technologies in Central and Eastern European (CEE) countries has also been documented in previous European reports [37,38].

Descriptive evidence shows that while AI adoption increased between 2021 and 2024, productivity developments did not follow a comparable direction. The absence of a consistent short-term relationship between AI usage and productivity is in line with the broader literature, which highlights that the economic benefits of AI typically materialise only after complementary organisational and managerial investments have been implemented [29,39,40]. Several studies emphasise that AI-related productivity gains require upgrades in processes, data infrastructure, and workforce skills, and therefore appear with a time delay rather than immediately [29,39].

In this context, the observed data do not indicate a clear short-term association between AI adoption and labour productivity at the aggregate level. The results further show that the largest adoption gaps occur in small and medium-sized enterprises, which face the most significant barriers to the diffusion of digital technologies. This aligns with international evidence indicating that SMEs lag behind larger firms in digital transformation and AI adoption due to resource constraints, limited expertise, and lower technological readiness [1].

The weak and inconsistent correlations between AI adoption and labour productivity observed in this study are also consistent with the so-called “productivity paradox” of artificial intelligence, where technological progress does not automatically translate into short-term productivity improvements [29]. Instead, productivity developments in Slovakia appear to be more heavily influenced by broader macroeconomic dynamics, shocks, and cyclical variations, which may overshadow any emerging effects of AI.

Overall, the findings suggest that while AI adoption is increasing, its observable short-term relationship with productivity remains limited without the presence of complementary factors such as organisational change, reskilling, and higher digital maturity. This supports the argument that AI-driven productivity gains require long-term structural adjustments rather than isolated technological uptake [29,40,41]. Moreover, different AI technologies may influence productivity through distinct channels; for example, generative AI, advanced analytics, and process automation can produce heterogeneous outcomes across firms and sectors.

From a policy perspective, the identified adoption gaps indicate that increasing AI uptake alone is unlikely to translate into measurable productivity improvements without complementary investments in skills, organisational capabilities, and innovation ecosystems. Targeted support for SMEs, diffusion of managerial know-how, and improved access to digital infrastructure, therefore, remain key priorities for enabling long-term economic benefits.

An important limitation of this study stems from the very small number of available observations and the short time horizon. With AI data limited to three survey years, any statistical relationships must be interpreted with extreme caution. The analysis, therefore, does not provide evidence of causal effects and should be understood strictly as descriptive and exploratory. Moreover, productivity responses to general-purpose technologies typically materialise over longer adjustment periods, often described in the literature as the J-curve effect. Consequently, the time span 2021–2024 is likely insufficient to capture the structural benefits of AI diffusion. In addition, sector-level heterogeneity could not be analysed due to data availability constraints in the harmonised dataset.

5. Conclusions

This study examined the adoption of artificial intelligence (AI) in Slovak enterprises compared with the EU27 and explored its association with labour productivity. The findings show that although AI adoption has increased, Slovakia remains below the EU27 average across all enterprise size classes, while labour productivity exhibits substantial volatility that does not mirror AI developments.

The results indicate no observable short-term association between AI adoption and productivity, and the illustrative regression does not indicate an immediate statistical relationship. Accordingly, the observed data do not provide evidence of such an association in the short term. Given that the benefits of general-purpose technologies frequently emerge only after longer adjustment horizons, the analysed period is too short to capture potential long-term productivity improvements.

This interpretation is consistent with previous research showing that the productivity effects of AI tend to emerge only over longer periods and require complementary investments in organisational change and skills development [40,41].

Although the analysis focuses on Slovakia, similar patterns may also be relevant for other small and medium-sized EU economies with comparable industrial structures, levels of digital maturity, and integration into European value chains. At the same time, institutional differences and sectoral compositions vary across countries, and therefore generalisation of the results should be made with appropriate caution.

Despite these limitations, the analysis points to considerations relevant for policymakers and enterprises. Improving digital skills, supporting complementary organisational innovation, and strengthening technological readiness—particularly among SMEs—may help enhance the long-term benefits of AI adoption. Future research should therefore rely on longer time series, firm-level microdata, and sector-specific indicators to better understand the mechanisms linking AI diffusion to economic performance and to enable more robust empirical inference.