AI upskilling helps keep stressed pharmaceutical workers on the job but doesn't immediately boost their performance; firms that deploy AI-guided training report smaller stress-related turnover even though stress still undermines output.

AI-driven stress management and performance optimization: A framework for employee retention in the IR 5.0 era

Devaraju, Amit Subramanyam · March 10, 2026

openalex correlational low evidence 7/10 relevance DOI Source PDF

In a cross-sectional survey of 350 Indian pharmaceutical workers, AI-enabled upskilling and AI-guided procedures attenuated the negative association between workplace stress and retention (but did not significantly moderate stress's effect on performance), while higher performance is positively associated with retention.

Stress in the modern workplace is a major factor affecting productivity and employee turnover. AI may reduce stress in pharmaceutical personnel, affecting their productivity and loyalty, according to this research. The study analyzes how AI improves productivity and reduces stress in everyday employment. Moreover, AI-facilitated upskilling initiatives enhance workers' confidence in their responsibilities, establishing AI as a vital intermediary among stress, performance, and retention. This quantitative research used purposive sampling to obtain data from 350 pharmaceutical workers in Karnataka, India. To examine the relationships between AIUP, stress, performance, retention, and SmartPLS 4.0 SEM. Research shows that stress negatively impacts performance (β = 0.158, p < 0.001) and retention (β = 0.321, p < 0.001). While AI procedures significantly controlled the relationship between stress and retention (β = 0.078, p < 0.005), their impact on stress and performance was not significant. Improved performance led to higher retention rates (β = 0.348, p < 0.001). The study shows AI's revolutionary potential in employment difficulties. AI promotes skill enhancement, increases work happiness, and aids in tailored stress management techniques, thereby boosting retention. These findings are in line with the human-centered technological advancements of IR 5.0 and demonstrate the vital role of AI in developing a long-term, adaptable workforce.

Summary

Main Finding

AI-enabled upskilling and AI-driven processes in Karnataka’s pharmaceutical sector reduce the negative effect of workplace stress on employee retention, increase retention directly, and support the Industry 5.0 goal of human-centered automation. Stress significantly lowers both performance and retention; performance strongly increases retention. However, AI upskilling (AIUP) significantly moderates the stress → retention link (b = 0.078, p = 0.005) but does not significantly moderate the stress → performance link (b = 0.044, p = 0.209).

Key Points

Core relationships (from SEM, SmartPLS 4.0):
- Stress → Performance: b = 0.158, T = 4.774, p < 0.001 (stress reduces performance).
- Stress → Retention: b = 0.321, T = 10.988, p < 0.001 (stress reduces retention).
- Performance → Retention: b = 0.348, T = 11.868, p < 0.001 (higher performance raises retention).
- AIUP → Retention (direct): b = 0.064, T = 2.269, p = 0.023 (AI processes modestly increase retention).
- Moderator effects:
  - AIUP x Stress → Retention: b = 0.078, T = 2.807, p = 0.005 (significant buffering effect).
  - AIUP x Stress → Performance: b = 0.044, T = 1.257, p = 0.209 (not significant).
Measurement quality:
- Cronbach’s α high for all constructs (AIUP 0.955; PER 0.934; RET 0.954; STR 0.937).
- AVE > 0.5 for constructs (AIUP 0.816; PER 0.788; RET 0.812; STR 0.796).
- Fornell-Larcker and HTMT show discriminant validity.
Predictive performance:
- Q2predict: PER = 0.628; RET = 0.523 (adequate predictive relevance).
- PLS Predict/CVPAT: model predicts RET significantly better than baseline (average loss difference negative and significant overall; RET p = 0.000), less clear improvement for PER.

Data & Methods

Design: Cross-sectional quantitative survey, purposive sampling.
Sample: 350 employees in Karnataka’s pharmaceutical cluster (manufacturing, quality testing, inventory management); balanced sector coverage; majority male (67%), largest age group 31–45 (44%), mixed AI familiarity (30% highly familiar, 44% somewhat).
Measurement/Analysis:
- Constructs: Stress (STR), Employee Performance (PER), Retention (RET), AI Upskilling/Processes (AIUP).
- Analysis: Structural Equation Modeling using SmartPLS 4.0; reliability (Cronbach’s α, rho_a, rho_c), convergent validity (AVE), discriminant validity (Fornell-Larcker, HTMT), predictive checks (Q2predict, PLS Predict/CVPAT).
Key limitations noted by study design (implicit):
- Purposive and sector-specific sample limits generalizability.
- Cross-sectional self-report survey — causality cannot be firmly established.

Implications for AI Economics

Labor productivity and retention economics:
- AI-driven upskilling and automation can reduce turnover by improving perceived support and matching skills to tasks; lower turnover translates to reduced hiring/training costs and preserved firm-specific human capital.
- The direct effect of AIUP on retention (though modest) and its buffering of stress → retention implies positive returns to investments in AI-enabled HR/training systems, especially where stress-driven attrition is high.
Wage, skill, and task composition:
- Findings support the complementary view of AI: when paired with upskilling, AI enhances worker resilience rather than simply substituting labor, suggesting policy emphasis on retraining to capture productivity gains without displacing workers.
Firm-level investment strategy:
- Firms should prioritize AI deployments that automate routine tasks and embed tailored upskilling/stress-management features; these generate retention benefits and likely reduce hidden costs of stress (absenteeism, presenteeism).
- Cost–benefit analyses should include turnover savings, productivity retention from experienced workers, and reduced stress-related performance losses—model results show meaningful effects on retention and predictive accuracy for retention outcomes.
Policy and broader economic considerations:
- Subsidies or tax incentives for AI upskilling programs could be economically justified as they increase labor market attachment and human capital durability in sectors with high stress exposure.
- Labour-market interventions should combine technology adoption with workforce development to realize distributional gains and mitigate displacement risks.
Research gaps for economists:
- Need for longitudinal and causal identification (e.g., difference-in-differences, randomized trials) to quantify causal returns to AI investments on retention and productivity.
- Heterogeneity analysis by job type, skill level, and AI familiarity to estimate distributional impacts and net welfare effects.
- Full accounting of costs (AI deployment, training) vs. benefits (reduced turnover, productivity) to produce ROI benchmarks for firms and policymakers.

Suggested next steps for applied economic work: estimate turnover cost savings from observed retention effects; run longitudinal trials tying AI-upskilling interventions to objective productivity metrics; and model equilibrium labor-market responses to scaled AI-upskilling adoption.

Assessment

Paper Typecorrelational Evidence Strengthlow — Cross-sectional, purposive survey with self-reported measures and no quasi-experimental variation or longitudinal design; therefore associations (including moderation) may reflect reverse causality, omitted variables, or common-method bias rather than causal effects. Methods Rigormedium — Analysis uses an appropriate multivariate technique (PLS-SEM) with an adequate sample size for that method and reports path coefficients and significance; however, sampling was non-probability, measurement and control-variable details are sparse in the provided summary, and key threats (endogeneity, common-method variance, robustness checks) are not addressed. SamplePurposive sample of N=350 pharmaceutical-sector employees in Karnataka, India; cross-sectional survey measuring workplace stress, self-reported job performance, retention intent/behavior, and AI-related upskilling/AI-guided procedures (AIUP); instrument/scale specifics and control variables not provided here. Themesskills_training human_ai_collab labor_markets GeneralizabilityGeographically limited to Karnataka, India — regional/contextual labor market and cultural factors may not generalize elsewhere, Sector-specific (pharmaceutical personnel) — findings may not apply to other industries or occupations, Purposive (non-probability) sampling limits population representativeness, Cross-sectional, self-reported measures raise concerns about common-method bias and reverse causality, Sample size moderate but not large enough to ensure external validity across diverse firm sizes and types

Claims (8)

Claim	Direction	Confidence	Outcome	Details
Workplace stress is associated with lower employee retention. Turnover	negative	high	employee retention (retention intent/behavior)	n=350 β = 0.321, p < 0.001 0.15
Workplace stress is associated with reduced job performance. Output Quality	negative	medium	job performance	n=350 β = 0.158, p < 0.001 0.09
Higher job performance is positively associated with greater employee retention. Turnover	positive	high	employee retention	n=350 β = 0.348, p < 0.001 0.15
AI-enabled upskilling and AI-guided procedures weaken the negative effect of workplace stress on employee retention (AI moderates the stress→retention link). Turnover	positive	medium	employee retention	n=350 β = 0.078, p < 0.005 0.09
AI did not significantly moderate the relationship between workplace stress and job performance. Output Quality	null_result	medium	job performance	n=350 0.09
The study used a cross-sectional quantitative survey (purposive sampling) of pharmaceutical-sector employees in Karnataka, India (N = 350) and analyzed relationships using PLS-SEM (SmartPLS 4.0). Other	null_result	high	study design / methodological characteristics	n=350 0.15
Because the design is cross-sectional and sampling purposive/geographically constrained, causal inference and generalizability are limited. Other	negative	high	generalizability / causal inference (methodological limitation)	n=350 0.15
Interpretive claim: AI interventions (upskilling and AI-guided workflows) raise worker confidence and job satisfaction and help tailor stress-management approaches, which can support retention under stress. Worker Satisfaction	positive	speculative	worker confidence / job satisfaction (interpretive mechanism for retention effects)	0.01