Employees frequently conceal AI use to avoid higher workloads or job risk, turning efficiency tools into sources of anxiety and mistrust; Gen Z both embraces AI for productivity and pursues careers perceived as less automatable, forcing firms to rethink management and performance metrics.

Main Finding

AI at work creates a paradox: although organisations promote AI as an efficiency and creativity booster, managerial practices and organisational culture frequently turn AI into a source of surveillance and pressure. That dynamic drives widespread concealment of AI use by employees, reduces psychological safety, and generates a “productivity trap” where efficiency gains are recycled into higher output expectations. Generation Z responds by both embracing AI as a “copilot” and seeking roles perceived as less automatable, forcing firms to rethink incentives, governance, and performance metrics if they want genuinely productive human–AI complementarities.

Key Points

• AI is increasingly embedded in managerial processes (algorithmic management), shifting supervision into automated monitoring and evaluation and extending a form of “technological Taylorism.”
• The “AI concealment effect”: many employees privately use AI but hide that use for fear of job loss, higher expectations, reputational damage, or managerial sanctions.
• Psychological drivers of concealment include impression management, AI anxiety (fear of replacement), technological shame, and low psychological safety.
• Organisational culture is decisive: learning-oriented, transparent cultures increase open AI use; control-oriented, hierarchical cultures produce “digital silence.”
• Generation Z is ambivalent: digitally fluent and pragmatic about AI’s productivity benefits, yet concerned about displacement and desirous of meaningful, non-automatable work (e.g., creative, empathetic, skilled trades).
• Empirical evidence cited (secondary sources): Deloitte (2025) and other surveys showing both optimism and fears among younger cohorts; Ransbotham et al. (2022) on business value and self-reported AI use; Resume Builder (2025) on Gen Z interest in blue-collar/trade roles.
• Policy and managerial framing matter: transparency, ethical governance, participatory training, and redefined performance metrics can make AI an augmenting partner rather than a disciplinary device.
• Limits: this is a conceptual literature review (no primary data); many claims rest on prior surveys and case studies, calling for dedicated empirical work.

Data & Methods

• Type: Conceptual paper / literature review synthesising academic studies, practitioner reports, and surveys rather than presenting original empirical data.
• Sources synthesised: peer‑reviewed research on algorithmic management and organisational psychology (e.g., Kellogg et al. 2020; Edmondson 2018; Burrell 2016), practitioner and survey reports (Deloitte 2025; McKinsey; Ransbotham et al. 2022; MIT Sloan / BCG), and sectoral case studies (PwC, IBM playbooks).
• Empirical evidence in the paper is secondary: selected statistics and survey results are used to illustrate patterns (e.g., workforce composition projections, percentages of Gen Z attitudes, rates of perceived AI benefits and fears).
• Methodological limitations noted by the authors: absence of primary data, potential sampling and reporting biases in cited surveys, and the need for causal and longitudinal studies to quantify concealment and its economic impacts.
• Recommended future empirical methods (implicit): firm-level case studies, matched employer-employee data, field experiments on managerial messaging and policy interventions, and measurement strategies to detect hidden AI use.

Implications for AI Economics

Practical and research implications relevant to labor economics, organisational economics, and policy:

• Measurement and bias

  • Hidden AI use biases empirical estimates of automation’s effects. If workers conceal AI, standard surveys and productivity measures may misattribute output changes to worker effort or firm practices.
  • Productivity gains that are not openly acknowledged complicate assessments of AI-driven capital–labor complementarities and mismeasure returns to skills.

• Labor supply, job composition, and wages

  • Generation Z’s preference for less automatable, dignity-preserving jobs may shift occupational supply (greater take-up of skilled trades, creative and care occupations), altering relative wages and labor shortages across sectors.
  • Concealment and fear of workload increases can increase turnover, reduce human capital accumulation in white-collar roles, and affect wage bargaining dynamics.

• Firm incentives and redistribution of gains

  • Management often reinvests efficiency gains into higher output expectations (the “productivity trap”), creating negative externalities (stress, lower retention). Economists should account for the welfare costs of intensified workload when computing AI’s net gains.
  • There is a case for institutional mechanisms (internal governance, bargaining, taxation, or mandated redistribution) to ensure efficiency gains improve worker well-being, not only firm margins.

• Human–AI complementarities and skill policy

  • Effective AI adoption depends on retooling performance metrics, training, and work design to reward augmented outcomes rather than visible effort.
  • Policy should support reskilling programs targeted at tasks that complement AI and protect entry-level training opportunities that AI might automate.

• Organizational design and regulation

  • Transparent governance, clear AI-use policies, and protection of psychological safety are not just HR preferences—they materially affect AI uptake and productivity. Regulatory attention to workplace algorithmic transparency and worker voice (e.g., disclosure, appeal rights) can improve allocative outcomes.
  • Surveillance-related negative externalities (lower morale, reduced creativity) justify considering limits on algorithmic monitoring or mandatory disclosures of automated evaluations.

• Research priorities for economists

  • Quantify the prevalence and economic impact of AI concealment (using audits, incentivised surveys, or digital trace methods).
  • Measure how managerial framing alters firm-level returns to AI (randomised messages or training interventions).
  • Estimate welfare trade-offs: productivity gains vs. mental health, turnover costs, and long-run human capital effects.
  • Study heterogeneous effects by generation, sector, and occupation to forecast labor-market reallocations and policy needs.

Recommended immediate policy/managerial actions (economics-informed) - Firms: adopt transparent AI governance, redesign performance metrics to credit outcomes (not visible effort), and invest in participatory training and psychological safety. - Policymakers: incentivise or mandate disclosures about algorithmic evaluation systems, support retraining for complementarity skills, and consider redistributive mechanisms for AI gains. - Researchers: build datasets linking firm AI-adoption, worker-reported use (including concealed use proxies), and objective outcomes (productivity, turnover, wages).

Overall, the paper argues that the economics of AI adoption cannot ignore organisational culture, incentives, and concealed behaviour; incorporating these social mechanisms is essential to accurately evaluate AI’s productivity, distributional consequences, and policy choices.