The Commonplace
Home Papers Evidence Explore Trends Syntheses Digests References Docs 🎲 Workforce Futures
← Papers
Direction, evidence grade, and study type are AI-generated labels (gpt-5-mini), not human-verified. Syntheses are LLM-written. "Tensions" are machine-detected candidates, not confirmed contradictions. A research-acceleration tool, not peer review. How this is built →

Generative AI both confuses and empowers workers: among 541 Chinese high‑tech employees, AI use raised role uncertainty (deterring collaboration) yet also boosted workers’ confidence to take on broader tasks (encouraging collaboration); stronger AI literacy dampened confusion and amplified confidence, and collaboration correlated with higher self-reported performance.

Dual pathways of generative AI use: role ambiguity and self-efficacy in employee-AI collaboration
Qiannan Zhang, Jingyi Zhang, Shan Dong · August 14, 2026 · Frontiers in Psychology
openalex correlational low evidence 7/10 relevance Full text usable extracted full text DOI Source PDF

Structured author observations

Linked only from stored provider relations; the raw author line above is never matched by name.

OpenAlex

Latest observation:

  1. Qiannan Zhang provider ID
  2. Jingyi Zhang provider ID
  3. Shan Dong provider ID

Semantic Scholar

Latest observation:

  1. Qiannan Zhang provider ID
  2. Jingyi Zhang provider ID
  3. ShanQiang Dong provider ID
In a survey of 541 employees at Chinese high-tech firms, generative AI use was associated with both higher role ambiguity (which reduced employee-AI collaboration) and higher role breadth self-efficacy (which increased collaboration), with AI literacy weakening the ambiguity link and strengthening the self-efficacy link, and collaboration tied to higher self-reported job performance.

Citation observations

Cumulative provider counts captured on specific dates; providers are never combined.

Introduction The growing use of generative AI tools in everyday work improves efficiency but also creates uncertainty about how employees interpret their roles and whether they are willing to collaborate with AI. Drawing on role theory, this study aims to investigate how generative AI (GenAI) use relates to employee-AI collaboration via two distinct pathways: role ambiguity as a hindrance mechanism and role breadth self-efficacy as an enabling mechanism. This study further examines the moderating role of AI literacy in the above relationships. Methods Survey data were collected from 541 employees working in Chinese high-technology firms, and structural equation modeling was adopted to test the proposed research model. Results The empirical results reveal that GenAI use simultaneously increases role ambiguity and role breadth self-efficacy. Role ambiguity negatively predicts employee-AI collaboration, whereas role breadth self-efficacy exerts a positive effect. AI literacy significantly moderates both pathways: it weakens the positive association between GenAI use and role ambiguity, and strengthens the positive link between GenAI use and role breadth self-efficacy. Furthermore, employee-AI collaboration is positively associated with job performance. Discussion These findings extend role theory by verifying that GenAI functions as both a disruptor and an enabler of employees’ role perceptions, and AI literacy serves as a vital boundary condition shaping how employees construct and understand their work roles when adopting GenAI. Practically, organizations are suggested to clarify work role boundaries, foster employees’ role breadth self-efficacy, and deliver systematic AI literacy training to facilitate high-quality human-AI collaboration.

Summary

Main Finding

Generative AI (GenAI) use affects employee-AI collaboration through two simultaneous, opposing role-based pathways. GenAI increases (1) role ambiguity, which reduces employees’ willingness to collaborate with AI, and (2) role breadth self-efficacy, which increases collaboration. Individual AI literacy moderates both pathways: higher AI literacy weakens the GenAI → role ambiguity link and strengthens the GenAI → role breadth self-efficacy link. Employee-AI collaboration is positively associated with job performance.

Key Points

  • Dual pathways:
    • Hindrance pathway: GenAI use → higher role ambiguity → lower employee-AI collaboration.
    • Enabling pathway: GenAI use → higher role breadth self-efficacy → higher employee-AI collaboration.
  • Moderation by AI literacy:
    • AI literacy attenuates the negative (role ambiguity) pathway.
    • AI literacy amplifies the positive (role breadth self-efficacy) pathway.
  • Outcome linkage: Greater employee-AI collaboration is associated with improved job performance.
  • The paper frames these effects using role theory (role ambiguity, role breadth self-efficacy) and social cognitive theory (sources of self-efficacy).
  • Practical takeaways emphasized by the authors: clarify role boundaries, foster role-breadth self-efficacy, and provide systematic AI literacy training to facilitate productive human-AI collaboration.

Data & Methods

  • Sample: Survey of 541 employees from Chinese high-technology firms.
  • Key constructs:
    • GenAI use (frequency/extent of using generative-AI tools),
    • Role ambiguity,
    • Role breadth self-efficacy,
    • AI literacy (ability to understand, interact with, evaluate AI),
    • Employee-AI collaboration (engagement in collaborative behaviors with AI),
    • Job performance (self-reported).
  • Design: Cross-sectional survey with validated scales (authors cite role/self-efficacy literatures for measures).
  • Analysis: Structural equation modeling to test hypothesized direct, indirect, and moderating relationships.
  • Main empirical outcomes: support for hypothesized dual pathways and the moderating role of AI literacy; positive association between collaboration and job performance.
  • Limitations (noted or implied): cross-sectional/self-report design limits causal inference; sample restricted to Chinese high-tech firms, so generalizability may be limited; future work suggested to use longitudinal, behavioral, or experimental designs.

Implications for AI Economics

  • Heterogeneous effects of AI adoption: GenAI does not uniformly substitute or complement labor; its net effect depends on psychological role responses and worker skills (AI literacy). Economic models should incorporate heterogeneity in worker interpretation and capability to use AI, not only task-level automatability.
  • Complementarity driven by skills and cognition: AI literacy shifts GenAI’s impact from disruptive (role ambiguity, withdrawal) toward complementary (role expansion, collaboration). Investment in AI literacy (training, onboarding) is likely to increase returns to AI adoption by converting potential displacement into productivity gains.
  • Labor reallocation and role redesign: GenAI can both blur and expand role boundaries. Firms that proactively redesign roles and clarify responsibilities may avoid costly withdrawal effects and realize productivity improvements. This suggests organizational policies (job redesign, incentives) matter for whether AI adoption raises firm-level labor productivity.
  • Distributional consequences: If AI literacy is uneven across workers (by education, tenure, occupation), GenAI adoption may increase within-firm inequality—those with higher AI literacy capture more complementarities (higher performance, expanded roles), while others face ambiguity and possibly lower engagement. This has implications for wage dynamics and skill premiums.
  • Empirical measurement recommendations for economists:
    • Include worker-level AI literacy and role-perception measures when estimating AI’s labor market effects.
    • Model interactions between technology exposure and human capital (especially AI literacy) to capture complementarity/substitution heterogeneity.
    • Use longitudinal designs or field experiments to separate causal effects of GenAI use from selection and reporting biases.
  • Policy implications: Public and firm-level training programs targeting AI literacy can increase aggregate gains from GenAI adoption and mitigate negative welfare or displacement effects. Regulation and guidance on role clarity, accountability, and job design may help preserve human agency and maximize productive collaboration.

If you want, I can extract likely questionnaire items/measures used for the constructs (GenAI use, role ambiguity, role-breadth self-efficacy, AI literacy, collaboration, performance) or propose an empirical specification for an economics study building on these findings.

Assessment

Paper Typecorrelational Evidence Strengthlow — Cross-sectional self-report survey data with SEM establishes associations but not causal effects; common-method bias, omitted confounders, and lack of temporal ordering limit causal inference. Methods Rigormedium — Sample size is adequate (n=541) and structural equation modeling is appropriate for testing the theorized mediation/moderation model, but reliance on cross-sectional, self-reported measures without objective outcomes or experimental/quasi-experimental identification reduces rigor. SampleCross-sectional survey of 541 employees employed at Chinese high-technology firms; measures include self-reported generative AI use, role ambiguity, role breadth self-efficacy, AI literacy, employee-AI collaboration, and self-reported job performance; analyzed using structural equation modeling. Themeshuman_ai_collab productivity skills_training GeneralizabilitySample limited to Chinese high-technology firms — may not generalize to other sectors or countries, Likely includes earlier adopters of GenAI; findings may not apply to non-users or late adopters, All measures are self-reported, including job performance — possible reporting bias, Cross-sectional design limits external validity for causal claims or longitudinal dynamics, Cultural or organizational norms in China may shape role perceptions differently than elsewhere

Claims (8)

ClaimDirectionOutcomeConfidence & EvidenceDetails
GenAI use simultaneously increases employees’ role ambiguity and role breadth self-efficacy. Task Allocation mixed Employees’ role ambiguity and role breadth self-efficacy
Reading fidelity high
Study strength medium
n=541
0.3
GenAI use is positively associated with employees’ role ambiguity. Task Allocation positive Role ambiguity
Reading fidelity high
Study strength medium
n=541
0.3
Role ambiguity negatively predicts employee-AI collaboration. Team Performance negative Employee-AI collaboration
Reading fidelity high
Study strength medium
n=541
0.3
GenAI use is positively associated with employees’ role breadth self-efficacy. Skill Acquisition positive Role breadth self-efficacy
Reading fidelity high
Study strength medium
n=541
0.3
Role breadth self-efficacy positively predicts employee-AI collaboration. Team Performance positive Employee-AI collaboration
Reading fidelity high
Study strength medium
n=541
0.3
AI literacy weakens the positive association between GenAI use and role ambiguity. Task Allocation negative The association between GenAI use and role ambiguity
Reading fidelity high
Study strength medium
n=541
0.3
AI literacy strengthens the positive association between GenAI use and role breadth self-efficacy. Skill Acquisition positive The association between GenAI use and role breadth self-efficacy
Reading fidelity high
Study strength medium
n=541
0.3
Employee-AI collaboration is positively associated with job performance. Organizational Efficiency positive Job performance
Reading fidelity high
Study strength medium
n=541
0.3

Notes