Impact of Artificial Intelligence on Organizational Performance

Prepared by the researche  : Tahani hallal ^– Doctoral School of Law , Political, Administrative and Economics Sciences, Lebanese Universsity , Lebanon

Democratic Arabic Center

Journal index of exploratory studies : Eighteenth Issue – June 2025

A Periodical International Journal published by the “Democratic Arab Center” Germany – Berlin

:To download the pdf version of the research papers, please visit the following link

https://democraticac.de/wp-content/uploads/2025/06/%D9%85%D8%AC%D9%84%D8%A9-%D9%85%D8%A4%D8%B4%D8%B1-%D9%84%D9%84%D8%AF%D8%B1%D8%A7%D8%B3%D8%A7%D8%AA-%D8%A7%D9%84%D8%A7%D8%B3%D8%AA%D8%B7%D9%84%D8%A7%D8%B9%D9%8A%D8%A9-%D8%A7%D9%84%D8%B9%D8%AF%D8%AF-%D8%A7%D9%84%D8%AB%D8%A7%D9%85%D9%86-%D8%B9%D8%B4%D8%B1-%D8%AD%D8%B2%D9%8A%D8%B1%D8%A7%D9%86-%E2%80%93-%D9%8A%D9%88%D9%86%D9%8A%D9%88-2025.pdf

Abstract

This study empirically investigates how artificial intelligence (AI) influences organizational performance in a sample of 25 Lebanese firms. Grounded in a positivist paradigm and using a cross-sectional survey design, we measure seven constructs—AI adoption level, technological capability, employee AI competence, innovation-oriented culture, digital leadership, AI-related R&D investment, and data quality—through validated scales. Aggregate scores are analyzed with SPSS via Pearson correlations and simple regressions.

The results are mixed. Only three variables exhibit a positive, statistically significant effect on performance: employee AI competence (β = 0.695, p < 0.001; R² = 0.483), innovation culture (β = 0.562, p = 0.003; R² = 0.316), and AI R&D investment (β = 0.579, p = 0.002; R² = 0.336). Together, they explain nearly half of the variance in organizational performance. Conversely, AI adoption level, technological capability, digital leadership, and data quality show no direct, significant impact (R² ≤ 0.081; p > 0.16).

These findings suggest that AI’s value is realized primarily through human skills, an entrepreneurial culture, and sustained R&D funding rather than mere tool deployment or infrastructure upgrades. Managers should therefore prioritize workforce upskilling, foster experimentation, and allocate long-term budgets to AI initiatives. Limitations include the small, cross-sectional sample; future research should employ larger, sector-specific panels and longitudinal designs to capture delayed performance effects.

1. Introduction

Artificial intelligence (AI) has become a major driver of organizational transformation. From automated production systems to customer-relationship-management tools, AI is reshaping work practices, accelerating decision-making processes, and enabling new business models to emerge.

Yet, despite this growing enthusiasm, the concrete ways in which AI affects organizational performance remain under-explored. Executives still wonder to what extent these new technologies create real added value.

Against this backdrop, the present study empirically assesses AI’s impact on organizational performance. Specifically, it examines how various dimensions—such as AI adoption level, employee capabilities, and data quality—shape that performance.

• Research question

How do the adoption and use of artificial intelligence affect firm performance?

• Objectives

• Identify the AI-related factors that influence organizational performance.
• Measure the impact of AI adoption on the various dimensions of performance.

• Research questions

• What is the relationship between a firm’s level of AI adoption and its performance?
• How do employee AI competencies and technological capabilities shape that relationship?
• In what ways do data quality and organizational culture influence performance outcomes?

1.4.  Literature Review

1.4.1.  Artificial Intelligence and Organizations

Artificial intelligence encompasses a range of technologies that enable systems to perform tasks typically requiring human intelligence (Russell, S. J., & Norvig, P. , 2010). Companies now deploy AI solutions across many functions—including process automation, customer service, predictive marketing, and preventive maintenance—reshaping operational models and value creation.

1.4.2.  Organizational Performance

Organizational performance is broadly defined as a firm’s ability to meet its strategic objectives (Kaplan, 1996),It is multidimensional, typically encompassing:

Financial performance (e.g., profitability, revenue growth)

Operational performance (e.g., efficiency, cycle time, quality)

Innovation capacity (e.g., new products, R&D output)

Customer satisfaction (e.g., loyalty, service ratings)

1.4.3.  Facteurs influençant la performance par l’IA

A growing body of research shows that a firm’s ability to translate AI investments into superior results depends on several organizational characteristics.  (Haefner, N., Wincent, J., Parida, V., & Gassmann, O., 2021); (Brynjolfsson, E., & McAfee, A. , 2017) highlight six factors in particular: the overall level of AI adoption, the firm’s technological capability, employees’ AI-related skills, digital leadership, data quality, and an innovation-oriented culture. In the present study, each construct is measured with a scale that has been validated in prior work: AI-adoption intensity (Ameen, A., et al. , 2024), IT capabilities (Wu, Y., & Chen, S., 2023), non-expert AI competence (Henkel, M. & Kleespies, M. , 2023) ,innovation culture (Martins, E. C., & Terblanche, F., 2003), digital leadership (Zeike, S., Bradbury, K., Lindert, L., & Pfaff, H. , 2019), AI-focused R&D investment (Cao, L., Li, Z., & Huang, J. , 2022), and data-quality dimensions (Wang, R. Y., & Strong, D. M. , 1996). Organizational performance itself is assessed through the four Balanced-Scorecard perspectives proposed by (Kaplan, R. S., & Norton, D. P. , 1996).

2. Methodological Approach

This study adopts a positivist paradigm, which assumes that an objective reality can be captured through deductive reasoning and statistical analysis. Consistent with recent methodological guidance—where positivism is deemed well-suited to questionnaire surveys and quantitative tests of causal relationships—an online questionnaire was administered to executives and senior managers of Lebanese firms, and the resulting data were analyzed with parametric tests to evaluate the hypotheses. This design meets the standards of methodological rigor and validity emphasized in recent management-science discussions of positivist research  (Ali, 2024).

• Population and Sample

Study focuses on a purposive sample of 25 Lebanese firms of varying sizes and across multiple industries, regardless of whether they have already implemented AI solutions.

• Variables and Hypotheses

Dependent variable: organizational performance.

Independent variables: (1) AI-adoption level, (2) technological capability, (3) employee AI competence, (4) innovation-oriented culture, (5) digital leadership, (6) AI-related R&D investment, and (7) data quality.

• Research hypotheses

• H1: A higher level of AI adoption has a positive effect on organizational performance.
• H2: Greater technological capability positively influences organizational performance.
• H3: Higher employee AI competence enhances organizational performance.
• H4: An innovation-oriented culture positively impacts organizational performance.
• H5: Strong digital leadership increases organizational performance.
• H6: Increased AI-related R&D investment improves organizational performance.
• H7: Better data quality fosters higher organizational performance.

• Conceptuel model

3. Expected Results and Analysis Plan

The data processing will be carried out using the SPSS software.

• Analysis of results

H1: The level of AI adoption has a positive effect on organizational performance.

Table 4:correlation adoption-performance

The value of r is -0.086, indicating a very weak correlation.

A negative aspect to consider is that as AI adoption increases, performance tends to decline, although the correlation is so weak that it is almost insignificant.

The level of statistical significance is determined by the p-value. In this case, with p = 0.683, which is well above 0.05, the result is considered not significant.

This dataset does not allow us to reject the null hypothesis (H0), which states that there is no correlation.

Hypothesis H1, suggesting that AI adoption improves performance, was not validated.

Table 5: model adoption-performance

Table 6:anova adoption -performance

Simple Regression: B = -0.128, β = -0.086, t = -0.41, p = 0.683 (N = 25) – AI adoption has no significant effect on performance.

Table 7: coefficient adoption-performance

Model Quality: F(1, 23) = 0.17, R² = 0.007, Adjusted R² = -0.036, Standard Error = 0.662 – less than 1% of the variance explained.

Table 8: collenearity adaptation-performance

No collinearity detected (VIF = 1.00); Hypothesis H1 rejected.

H2: Technological capability positively influences performance.

Table 9:correlations capacity-performance

Correlation r = -0.160, p = 0.445 (N = 25): A weak and non-significant relationship between technological capability and performance (≈ 2.6% of variance explained).

The negative sign suggests that greater technological capability is not associated with better performance in this sample, but the effect is too weak to be interpreted.

Hypothesis H2 rejected: Technological capability does not appear to be a significant predictor of organizational performance.

Table 10:model capacity-performance

Simple Correlation: r = -0.160, p = 0.445.

– a very weak and non-significant relationship.

Table 11: anova capacity-performance

Simple Regression: B = -0.300, β = -0.160, t = -0.78, p = 0.445; F(1, 23) = 0.60, R² = 0.026 (Adjusted R² = -0.017), Standard Error = 0.656; no collinearity issues (VIF = 1.00).

Table 12: coefficient capacity-performance

Table 13: collinearity capacity-performance

Technological capability neither explains nor significantly predicts performance (< 3% of variance), with a very weak negative effect; H2 rejected.

H3 : Employee AI skills improve performance.

Table 14:correlations competence-performance

Table 15: model competence-performance

Table 16: anova competence-performance

Table 17:coefficinet competence-performance

Simple Regression: B = 1.142, β = 0.695, t = 4.63, p < 0.001

– AI skills have a strong and significant positive effect on performance.

Table 18: collinearity competence-performance

Model Quality: F(1, 23) = 21.45, R² = 0.483 (Adjusted R² = 0.460), SE = 0.478; nearly 48% of the variance in performance is explained.

No collinearity (VIF = 1.00).

Hypothesis H3 confirmed.

H4: Innovation-oriented culture has a positive impact.

Table 19: correlations culture -performance

Table 20: model culture -performance

Table 21:anoav:culture -performance

Model Quality: F(1, 23) = 10.61, R² = 0.316 (Adjusted R² = 0.286), SE = 0.550; about 32% of the variance in performance is explained.

Table 22:coefficient:culture -performance

Simple Regression: B = 1.145, β = 0.562, t = 3.26, p = 0.003

 – Innovation , oriented culture has a clear, positive, and significant effect on performance.

Table 23: collinearity culture -performance

No collinearity (VIF = 1.00); Hypothesis H4 confirmed.

H5: Digital leadership increases performance.

Table 24: correlations leadership -performance

 Correlation r = 0.285, p = 0.167: A positive but weak and non-significant relationship between digital leadership and organizational performance.

Explained variance: r² ≈ 0.08 → only about 8% of performance is related to leadership, a modest effect and statistically insufficient (α = 0.05).

Hypothesis H5 rejected: In this sample, digital leadership does not reliably predict performance. Further analysis on a larger sample or controlling for other variables may clarify this link.

Table 25:model: culture -performance

Model Quality: F(1, 23) = 2.04, R² = 0.081 (Adjusted R² = 0.041), Standard Error = 0.637

– only about 8% of the variance explained, overall model not significant.

Table 26: collinearity:culture -performance

Table 27: anova:culture -performance

Table 28: coefficients culture -performance

Simple Regression: B = 0.254, β = 0.285, t = 1.43, p = 0.167 – Digital leadership has no significant effect on performance.

No collinearity (VIF = 1.00); Hypothesis H5 rejected.

H6: Investment in AI R&D improves performance.

Table 29: correlations R&D-performance

Correlation r = 0.579, p = 0.002: A moderate to strong and highly significant positive relationship between AI R&D investment and organizational performance.

Table 30: model: R&D-performance

Model Quality: F(1, 23) = 11.62; R² = 0.336 (Adjusted R² = 0.307); SE = 0.542

 – approximately 34% of the variance in performance is explained, a notable and robust effect.

Table 31: anova R&D-performance

Table 32: coefficients :R&D-performance

Simple Regression: B = 0.630; β = 0.579; t = 3.41; p = 0.002 – AI R&D investment positively and significantly influences performance.

Table 33: collinearity R&D-performance

Table 34: correlation:R&D-performance

Correlation r = 0.263, p = 0.204 (N = 25): A positive but weak and non-significant relationship between data quality and performance.

Explained variance: r² ≈ 0.07 → only about 7% of performance is associated with data quality; the effect is too modest to draw a conclusion.

Hypothesis H7 rejected: In this sample, improving data quality does not appear to be a statistically reliable predictor of organizational performance

Table 35: model: R&D-performance

Model Quality: F(1, 23) = 1.71; R² = 0.069 (Adjusted R² = 0.029); SE = 0.641 – only 7% of the variance explained; overall model not significant.

Table 36: anova: R&D-performance

Table 37:coefficients R&D-performance

Simple Regression: B = 0.217; β = 0.263; t = 1.31; p = 0.204 – Data quality has no significant effect on performance.

Table 38: collinearity R&D-performance

No collinearity (VIF = 1.00). Hypothesis H7 rejected: In this sample, improving data quality does not appear to be a reliable predictor of organizational performance.

4. Discussion

Empirical tests reveal a nuanced picture of the impact of artificial intelligence (AI) on organizational performance. Out of the seven proposed hypotheses, three are confirmed (H3, H4, H6) and four are rejected (H1, H2, H5, H7). This heterogeneity suggests that not all dimensions of AI are equally impactful: some create measurable value, while others have only a marginal or even null effect in the sample studied.

H3 – AI Skills (β = 0.695, p < 0.001) stands out as the most decisive factor, explaining nearly 48% of the variance in performance. This result confirms the literature that states digital human capital plays a key role in capturing technological value: when employees can understand, configure, and leverage AI tools, productivity and innovation gains quickly materialize.

H4 – Innovation-Oriented Culture (β = 0.562, p = 0.003) and H6 – AI R&D Investment (β = 0.579, p = 0.002) explain 32% and 34% of performance, respectively. They emphasize the importance of an organizational environment open to experimentation and sustained financial commitment: AI not only requires development budgets but also a tolerance for risk and failure to generate new ideas and continuous improvements.

On the other hand, H1 – AI Adoption Level (β = -0.086, p = 0.683) and H2 – Technological Capability (β = -0.160, p = 0.445) are not significant. These results align with studies that highlight the “productivity paradox”: deploying AI solutions or having modern infrastructure does not automatically guarantee a return on investment. Without the right skills, a favorable culture, and R&D budgets, AI may remain underutilized or misaligned with business strategy.

H5 – Digital Leadership (β = 0.285, p = 0.167) does not reach the significance threshold. Two possible explanations: first, management may delegate AI implementation to technical teams, diluting the direct effect of leadership; second, the limited sample (N = 25) may obscure a real but moderate effect. Finally, H7 – Data Quality (β = 0.263, p = 0.204) does not appear decisive, suggesting that companies only fully leverage “good” datasets once the skills and processes are in place – a point to be explored in future longitudinal research.

5. Managerial Implications:

• Train before deploying: Investing in upskilling teams (AI training, re-skilling programs) appears more cost-effective than simply multiplying tools.
• Embed AI in the culture: Encouraging experimentation, valuing learning failures, and establishing innovation rituals foster the widespread adoption of AI practices.
• Budget AI R&D strategically: Financial resources should support pilot projects, proofs of concept, and scaling up.
• Monitor data governance: Although data quality is not significant here, it remains a prerequisite when skills and culture reach maturity.
  6. Limitations and Future Directions

The study relies on a small and cross-sectional sample; analyses with a larger, sector-specific, and longitudinal panel would allow for validating the robustness of the effects and exploring temporal dynamics (delayed effects of AI). Additionally, variables measured using item averages could be refined by using objective indicators (ROI, cycle time, defect rate). Finally, testing moderated or mediated models — such as “Technological Capability × Skills” — would shed light on the conditions under which AI fully realizes its potential.

7. Conclusion

This research provides a contrasting perspective on how artificial intelligence shapes organizational performance. Out of the seven relationships tested, only three — AI skills, innovation culture, and AI R&D investment — emerge as decisive levers, together explaining nearly half of the measured performance. In contrast, the mere level of adoption of tools, technological capability, digital leadership, and, to a lesser extent, data quality, show no significant direct effect in our sample.

These results confirm that a people-and-process-driven approach takes precedence over a purely technological one: without trained talent, a culture conducive to experimentation, and sustained R&D budgets, AI remains an underutilized potential. Decision-makers would therefore benefit from focusing their efforts on upskilling teams, embedding a learning culture, and strategically allocating resources rather than multiplying tools or solely modernizing infrastructure.

However, the study does have limitations — sample size, cross-sectional approach — which warrant caution and open the door to future work, including longitudinal and sector-specific analyses, as well as exploring moderating effects (e.g., company size or sector). Despite these reservations, our findings clearly highlight that in AI, value arises less from the technologies themselves than from an organization’s ability to learn, innovate, and invest for the long term.

Bibliography

Ali, I. M. (2024). A Guide for Positivist Research Paradigm: From Philosophy to Methodology. Idealogy Journal, 187-196.

Ameen, A., et al. . (2024). AI adoption intensity and sustainability performance, Sustainability. The Moderated Mediation Effect of Organizational Change, 16-21.

Brynjolfsson, E., & McAfee, A. . (2017). Machine, Platform, Crowd:: Harnessing the digital revolution. New York, NY: Norton & Company.

Cao, L., Li, Z., & Huang, J. . (2022). The role of R&D expenditure in technological innovation: Evidence from Chinese manufacturing firms. Technology in Society, 68.

Haefner, N., Wincent, J., Parida, V., & Gassmann, O. (2021). Artificial Intelligence and Innovation: A Review, framework, and research agenda. Technological Forecasting and Social Change, 162.

Henkel, M. & Kleespies, M. . (2023). Scale for the assessment of non-experts’ AI competence. Scale for the assessment of non-experts’ AI competence, 9(4).

Kaplan, R. S. (1996). The Balanced Scorecard.

Kaplan, R. S., & Norton, D. P. . (1996). The balanced scorecard: Translating strategy into action. Boston, MA : Harvard Business School Press.

Martins, E. C., & Terblanche, F. (2003). Building organisational culture that stimulates creativity and innovation. European Journal of Innovation Management64-74, 7(1), 64-74.

Russell, S. J., & Norvig, P. . (2010). Artificial intelligence: A modern approach . ((. éd.), Ed.) Upper Saddle River, NJ : Prentice Hall.

Wang, R. Y., & Strong, D. M. . (1996). Beyond accuracy: What data quality means to data consumers. Journal of Management Information Systems, 12(4), 5-34.

Wu, Y., & Chen, S. (2023). IT capabilities and organizational ambidexterity. Technology in Society, 75.

Zeike, S., Bradbury, K., Lindert, L., & Pfaff, H. . (2019). Digital leadership skills and associations with psychological well-being. International Journal of Environmental Research and Public Health, 16(14).