Employees who proactively expand their roles to work with generative AI report higher career satisfaction and better supervisor-rated performance because AI increases work meaningfulness, while defensive avoidance of AI erodes careers by increasing work alienation; autonomy magnifies the upside and blunts the downside, and these effects are confined to work rather than life satisfaction.

Main Finding

Employees respond to generative AI with two distinct proactive strategies—AI approach job crafting (leveraging AI to expand work and capabilities) and AI avoidance job crafting (contracting or distancing from AI). These produce divergent career outcomes via separate psychological pathways: approach crafting increases career-relevant outcomes (career satisfaction and leader-rated job performance) through greater work meaningfulness, whereas avoidance crafting decreases those outcomes via increased work alienation. Job autonomy amplifies the benefits of approach crafting and dampens the harms of avoidance crafting. Neither pathway affected general life satisfaction, indicating domain specificity. (Liu, Lei & Qu, Frontiers in Psychology, 2026; DOI: 10.3389/fpsyg.2026.1779227)

Key Points

• Conceptual contribution
  • Introduces "AI job crafting" as a contextualized extension of job crafting theory focused on generative AI (GenAI, LLM-driven tools).
  • Distinguishes two motivationally driven dimensions: AI approach job crafting (mastery/growth orientation) and AI avoidance job crafting (security/protection orientation).
  • Integrates approach-avoidance motivation with Job Demands–Resources (JD-R) to explain human–GenAI interaction.
• Mechanisms
  • Gain path: AI approach → increased work meaningfulness → higher career satisfaction and performance.
  • Loss path: AI avoidance → increased work alienation → lower career satisfaction and performance.
  • Work meaningfulness and alienation are treated as identity-level mediators capturing existential reactions to AI, not just task-level effects.
• Boundary conditions
  • Job autonomy moderates effects: high autonomy strengthens the positive approach→meaningfulness link and weakens the avoidance→alienation link.
• Domain specificity
  • Effects concentrate on proximal professional outcomes (career satisfaction, performance); no significant influence on life satisfaction was found.
• Measurement
  • Authors developed and validated an AI Job Crafting Scale tailored to GenAI contexts.

Data & Methods

• Design: Multi-source, multi-wave survey of employee–leader dyads.
• Sample: 287 dyads in China.
• Measures:
  • Newly developed AI Job Crafting Scale (subscales: approach and avoidance).
  • Mediators: work meaningfulness and work alienation.
  • Outcomes: career satisfaction and leader-rated performance; life satisfaction as a distal outcome.
  • Moderator: perceived job autonomy.
• Analysis:
  • Mediation models testing dual-path (gain and loss) indirect effects.
  • Interaction tests for job autonomy moderation.
  • Multi-wave and multi-source design to reduce common-method bias and strengthen causal inference (details on wave timing and exact estimation techniques reported in paper).
• Key statistical results:
  • Significant positive indirect effect: AI approach → meaningfulness → career satisfaction & performance.
  • Significant negative indirect effect: AI avoidance → alienation → career satisfaction & performance.
  • Moderation by job autonomy: enhances approach benefits; attenuates avoidance harms.
  • No significant indirect effects on life satisfaction.

Implications for AI Economics

• Microbehavioral heterogeneity matters for labor outcomes
  • Economists modeling AI impacts should treat worker responses to AI as endogenous and heterogeneous: some workers actively recompose tasks to complement AI (increasing productivity and career returns), while others withdraw (reducing productivity and career prospects).
  • Incorporate binary/continuous worker strategies (approach vs. avoidance) into models of human–AI complementarity and substitutability rather than assuming uniform adaptation.
• Human capital and upskilling dynamics
  • Approach crafting resembles investment in complementary human capital (learning to use AI to expand scope/complexity of work). This supports models where returns to upskilling increase under accessible AI and supportive environments.
  • Avoidance crafting resembles defensive preservation of existing human capital; it may slow reallocation and reduce wage growth potential, impacting occupational mobility models.
• Role of organizational institutions and policy
  • Job autonomy plays a systemic role: organizational design (task assignment, discretion, managerial use of AI) shapes whether AI integration yields aggregate productivity gains or creates alienation and stagnation.
  • Policy levers (workplace governance, retraining subsidies, regulation of algorithmic management) that increase worker discretion could raise the aggregate fraction of workers who adopt approach strategies.
• Measurement and empirical strategy recommendations
  • Use domain-specific instruments (e.g., the paper’s AI Job Crafting Scale) when studying AI effects on labor outcomes—generic measures of technology adoption may miss critical motivational dynamics.
  • Prefer multi-wave, multi-source designs (employee reports + supervisor outcomes) to identify psychological mediation and performance effects.
  • Include mediators such as work meaningfulness and alienation when estimating the effect of AI exposure on labor market outcomes; proximate career metrics may move before broad wellbeing measures.
• Aggregate and distributional consequences
  • If approach crafting is more accessible to higher-skill or higher-autonomy workers, AI adoption may amplify inequality (skill-premium effects and occupational polarization). Models of inequality should allow for sorting by autonomy and relative ability to craft jobs.
  • Conversely, improving firm-level autonomy and training could shift composition toward complementary strategies, raising aggregate productivity and preserving career sustainability.
• Modeling suggestions
  • Extend search-and-matching or human-capital accumulation frameworks to include an endogenous "crafting" choice that changes task scope and productivity returns; let organizational parameters (autonomy) alter the costs/benefits of crafting.
  • Incorporate psychological state variables (meaningfulness, alienation) as intermediates that affect labor supply decisions, effort, turnover, and investment in skills.
• Future empirical targets
  • Estimate how prevalence of approach vs. avoidance crafting mediates regional/national productivity responses to AI diffusion.
  • Test heterogeneity by occupation, firm size, regulatory environment, and initial skill level.
  • Experimentally or quasi-experimentally assess interventions that increase autonomy or reduce identity threat (e.g., training emphasizing augmentation) and measure shifts in crafting and performance.

References - Liu Y., Lei H., Qu X. (2026). Approach or avoidance? A dual-pathway model of job crafting in response to generative AI and its impact on career sustainability. Frontiers in Psychology, 17:1779227. DOI: 10.3389/fpsyg.2026.1779227

If you want, I can extract the AI Job Crafting Scale items and suggest how to operationalize them in surveys or incorporate the findings into a structural economic model.