← Papers

Moderate AI adoption calms workers, heavy deployment unnerves them: employees report lower job insecurity at intermediate levels of AI use but higher insecurity when AI is extensive; strong self-belief and transformational leadership blunt both effects.

The impact of artificial intelligence application on employees' job insecurity: the moderating roles of self-efficacy and transformational leadership

Wenxiu Fu, Hui Zhang · June 03, 2026 · Frontiers in Psychology

openalex correlational low evidence 7/10 relevance DOI Source PDF

The study finds a U-shaped relationship between AI application intensity and employees' job insecurity — moderate AI use reduces insecurity while excessive AI intensifies it — and shows that self-efficacy and transformational leadership buffer these effects.

Introduction Against the backdrop of the intelligent era, the widespread application of artificial intelligence (AI) has fundamentally reshaped the internal and external developmental ecosystems of organizations, exerting a profound impact on employees' work-related psychological states. Drawing on the Conservation of Resources Theory and the Cognitive Appraisal Theory of Stress, this study empirically explores the underlying mechanism through which AI application influences employees' job insecurity. Methods Data for this study were collected using a mixed online and offline distribution method, with all measures administered through employee self-reported questionnaires. A total of 453 questionnaires were distributed, including 242 online and 211 offline. We received 449 questionnaires, with 242 from the online channel and 207 offline. Following a stringent validity screening process conducted by the research team, 411 valid questionnaires were retained for analysis (219 online and 192 offline), resulting in an effective response rate of 90.73%. Results There is a significant U-shaped relationship between AI application and employees' job insecurity: moderate AI application reduces insecurity, whereas excessive application heightens it. Self-efficacy negatively moderates this relationship by strengthening the insecurity-reducing effect of moderate AI application and weakening the insecurity-enhancing effect of excessive application. Transformational leadership similarly exerts a negative moderating effect, suggesting that both individual psychological resources and supportive leadership can buffer employees' insecurity responses to varying levels of AI application in digitality transforming organizational contexts more effectively. Discussion This study advances research on AI-enabled workplace changes by revealing a U-shaped effect of AI application intensity on employees' job insecurity. It explains this relationship through the dual mechanisms of resource gain and resource threat. It further incorporates self-efficacy and transformational leadership as boundary conditions, thereby clarifying when AI application alleviates or intensifies employees' job insecurity. These findings enrich the theoretical understanding of employees' job insecurity within the context of AI application, and offer empirical insights for managing employee wellbeing and refining human resource strategies during organizational digital transformation.

Summary

Main Finding

The study finds a U-shaped relationship between organizational AI application intensity and employees’ job insecurity: moderate AI use tends to reduce job insecurity (via resource gains and task complementarities), while excessive AI deployment increases insecurity (via substitution and resource threat). Two buffers—employee self-efficacy and transformational leadership—attenuate that U-shape: high self-efficacy and stronger transformational leadership both strengthen the insecurity-reducing effect at moderate AI levels and weaken the insecurity-increasing effect at high AI levels.

Key Points

Theoretical framing: Conservation of Resources (COR) Theory explains resource gains vs. threats from AI; Cognitive Appraisal Theory of Stress explains how self-efficacy and leadership shape employees’ appraisals.
Hypotheses tested: (H1) a positive U-shaped AI–job insecurity relationship; (H2) self-efficacy negatively moderates this U-shape; (H3) transformational leadership negatively moderates the U-shape.
Empirical result summary:
- Low → moderate AI adoption: innovation/complementarity effects dominate → lower job insecurity.
- High AI adoption: substitution/skill-depreciation effects dominate → higher job insecurity.
- High self-efficacy and high transformational leadership buffer employees from insecurity across AI intensities.
Practical interpretation: psychological resources (self-efficacy) and supportive leadership reduce perceived resource loss and technological threat.

Data & Methods

Study type: Cross-sectional original research (open access; Frontiers in Psychology).
Sample: Mixed online and offline employee survey in China. 453 questionnaires distributed (242 online, 211 offline); 449 returned; 411 valid after screening (219 online, 192 offline). Effective response rate reported: 90.73%.
Measures: Self-report scales for AI application intensity, employees’ job insecurity, self-efficacy, and transformational leadership (standard psychometric scales implied).
Analysis: Regression models including quadratic term for AI application to detect U-shaped relationship; interaction terms for moderators (self-efficacy and transformational leadership). Findings consistent with hypothesized U-shape and negative moderation by the two moderators.
Limitations noted or implied: cross-sectional self-report data (limits causal inference), potential common-method bias, sample context (Chinese organizations) may limit generalizability.

Implications for AI Economics

Adoption intensity matters for labor market welfare: moderate AI adoption can be complementary to labor (raising productivity and reducing perceived insecurity), while aggressive substitutional adoption can raise perceived and real job insecurity—potentially increasing transitional unemployment, wage pressure in vulnerable occupations, and demand for social insurance.
Policy and firm strategy:
- Pace AI rollouts to capture complementarities before large-scale substitution; staggered adoption reduces immediate displacement risk.
- Invest in human capital (reskilling/upskilling) and certification programs to reduce skill depreciation and ease labor reallocation.
- Support interventions that raise worker self-efficacy (training design emphasizing mastery experiences, coaching) to improve adaptive responses to AI.
- Encourage leadership development (transformational leadership practices) to reduce uncertainty and increase acceptance of technology-driven change.
- Consider targeted labor-market policies (retraining subsidies, portable benefits, wage insurance) in sectors prone to high-penetration AI.
Research implications for AI economics:
- Models of AI-driven productivity should endogenize adoption intensity and account for non-linear labor effects (complementarities at moderate adoption vs. substitution at high adoption).
- Incorporate behavioral and organizational moderators (worker beliefs, managerial styles) when forecasting displacement, wage dynamics, and labor supply responses.
- Need for longitudinal and quasi-experimental studies to identify causal impacts of differing AI adoption paths on employment, wages, and inequality.

Assessment

Paper Typecorrelational Evidence Strengthlow — Cross-sectional, self-reported survey data with no plausibly exogenous variation in AI adoption; potential reverse causality, omitted variable bias, and common-method variance undermine causal claims despite statistical controls. Methods Rigorlow — Moderate sample size (n=411) and use of quadratic and moderation analyses are appropriate for detecting associations, but reliance on convenience sampling, single-source self-reports, limited information on measurement validity and sampling frame, and absence of longitudinal or experimental design limit methodological rigor. Sample411 employees (219 online, 192 offline) surveyed via mixed online/offline convenience sampling; measures are all self-reported questionnaire items; detailed demographics, industries, country/organization sampling frame, and objective measures of AI use are not reported in the summary. Themeslabor_markets human_ai_collab org_design adoption GeneralizabilityConvenience and mixed self-selected sampling limits representativeness to broader worker populations, Unclear geographic, industry, or organizational coverage — findings may be context/culture-specific, All measures are self-reported, raising concerns about common-method bias and social desirability, Cross-sectional design prevents generalization about causal dynamics over time, AI application is measured subjectively (self-report intensity) rather than objectively validated deployment metrics

Claims (6)

Claim	Direction	Confidence	Outcome	Details
There is a significant U-shaped relationship between AI application and employees' job insecurity: moderate AI application reduces insecurity, whereas excessive application heightens it. Worker Satisfaction	mixed	high	job insecurity	n=411 0.3
Self-efficacy negatively moderates the relationship between AI application and employees' job insecurity by strengthening the insecurity-reducing effect of moderate AI application and weakening the insecurity-enhancing effect of excessive application. Worker Satisfaction	negative	high	job insecurity	n=411 0.3
Transformational leadership negatively moderates the relationship between AI application and employees' job insecurity, buffering employees' insecurity responses across varying levels of AI application. Worker Satisfaction	negative	high	job insecurity	n=411 0.3
Data were collected via mixed online and offline questionnaires: 453 questionnaires were distributed (242 online, 211 offline); 449 were returned (242 online, 207 offline); following validity screening, 411 valid questionnaires were retained (219 online, 192 offline), yielding an effective response rate of 90.73%. Other	null_result	high	survey response / valid sample size / response rate	n=411 effective response rate 90.73% 0.5
The study draws on the Conservation of Resources Theory and the Cognitive Appraisal Theory of Stress to explain how AI application influences employees' job insecurity via resource gain and resource threat mechanisms. Other	null_result	high	theoretical explanation of mechanisms behind job insecurity	0.05
The findings provide empirical insights for managing employee wellbeing and refining human resource strategies during organizational digital transformation. Other	null_result	high	managerial implications for employee wellbeing and HR strategies	n=411 0.05