Firms that cultivate psychological safety and organizational support dramatically raise employee uptake of collaborative AI by turning anxiety about automation into opportunities for learning and augmentation; in contrast, competitive or unsupportive climates amplify technostress, resistance and turnover risk.

Main Finding

A supportive organizational psychological climate — especially psychological safety, perceived organizational support, trust in automation, and a learning orientation — fundamentally shapes whether workers perceive collaborative AI as a threatening job demand or an empowering job resource. Using a PRISMA-guided systematic review of 92 studies (2015–2026), the authors synthesize evidence and propose the Psychological Climate‑Mediated AI Acceptance Model (PC‑MAAM): psychological climate mediates and moderates cognitive appraisals and behavioral intentions toward human–AI collaboration, thereby determining adoption, engagement, and well‑being outcomes.

Key Points

• Scope and approach

  • Systematic literature review following PRISMA of 92 peer‑reviewed empirical and theoretical studies (2015–2026).
  • Thematic synthesis integrating organizational behavior insights with information‑systems acceptance research.

• Theoretical framing

  • Integrates Job Demands‑Resources (JD‑R) theory and Conservation of Resources (COR) theory.
  • Psychological climate acts as a macro‑resource that shifts appraisal and resource investment dynamics under COR and JD‑R logics.

• Psychological climate components highlighted

  • Psychological safety (ability to experiment, admit errors without penalty).
  • Perceived organizational support (POS).
  • Trust in automation and leadership.
  • Learning orientation / growth mindset.

• Dual nature of AI at work

  • AI as job demand: increased cognitive load, learning anxiety, algorithmic monitoring, role ambiguity, technostress.
  • AI as job resource: cognitive augmentation, task variety, automation of repetitive work, potential for higher‑value tasks.

• Individual moderators

  • AI self‑efficacy, learning goal orientation, tenure/experience (seniority can increase resistance absent tailored support).

• Empirical patterns

  • Supportive climates correlate with greater initial uptake, exploratory use, and long‑run engagement with AI.
  • Competitive or unsupportive climates amplify anxiety, resistance, knowledge hiding, and turnover intentions.
  • Acceptance cannot be reduced to PU/PEOU alone; trust, perceived vulnerability, and relational factors are critical.

Data & Methods

• Method: Systematic literature review compliant with PRISMA; thematic synthesis of 92 peer‑reviewed articles (both empirical and theoretical) published between 2015 and 2026.
• Evidence types: cross‑sectional surveys, firm‑level case studies, experiments, meta‑analyses, and conceptual work spanning organizational behavior, information systems, and human–AI interaction literatures.
• Analytical output: synthesis of themes + integrated conceptual framework (PC‑MAAM) linking psychological climate, JD‑R/COR mechanisms, individual moderators, and adoption outcomes.
• Limitations noted by authors: predominantly non‑experimental evidence base, heterogeneity in measurement of psychological climate and AI constructs, need for longitudinal and causal studies.

Implications for AI Economics

• Adoption and returns-to-AI
  • Organizational internalities (psychological climate) materially affect uptake rates and effective utilization of AI, hence altering realized productivity gains and ROI on AI investments. Standard adoption models and cost‑benefit analyses should incorporate firm‑level climate measures as key determinants of realized returns.
• Complementarities and labor demand
  • Psychological climate modulates human–AI complementarities: supportive climates make AI more likely to augment human tasks (raising skilled labor productivity), while unsupportive climates increase substitution risks and negative labor outcomes.
• Human capital investment and wages
  • Heterogeneous adoption driven by climate and individual moderators implies differential returns to reskilling investments across workers and firms. Economics models should allow for firm‑level heterogeneity in the rate of human capital depreciation/gain when AI is introduced.
• Aggregate and distributional effects
  • Because psychological climate varies systematically across sectors and firm types, micro‑level climate effects can amplify inequality: firms with better climates capture more productivity gains, widening firm‑level and worker‑level dispersion.
• Policy design
  • Policies that aim to maximize social returns from AI should go beyond subsidizing technology and include support for organizational practices: funding for reskilling, incentives/standards for transparent algorithmic management, and grants for interventions that improve psychological safety and POS.
• Empirical research directions for economists
  • Measure and include psychological climate variables in firm‑level panels and worker surveys when estimating productivity effects of AI.
  • Use quasi‑experimental and randomized interventions (e.g., randomized managerial training, team‑level psychological safety interventions) to identify causal impacts on AI adoption and productivity.
  • Model endogenous investment in climate (organizational practices) as part of firms' optimization over technology adoption and human capital.
  • Estimate distributional impacts by linking climate‑mediated adoption to wages, employment transitions, and turnover across worker skill groups.
• Practical implications for firms and practitioners
  • To maximize adoption benefits, investments should bundle technology with organizational interventions (leadership training, transparent AI governance, learning programs) — otherwise capital expenditures risk underperformance due to human‑side frictions.

Recommended immediate steps for empirical AI economists - Incorporate standardized psychological climate scales into datasets used for evaluating AI impacts. - Treat psychological climate as both mediator and moderator in causal frameworks. - Design field experiments testing whether climate interventions increase effective AI use and productivity, and quantify cost‑effectiveness relative to pure technology subsidies.