Agentic AI is set to sharpen labor-market divides: strategic and creative professionals gain outsized productivity boosts while many middle-skill analytical and administrative roles face heightened displacement risk, and the transition is likely to unfold faster than past technology waves.

Main Finding

Agentic AI—systems that autonomously reason, plan, and execute multi-step tasks while integrating with enterprise tools—will accelerate labor-market polarization faster than prior automation waves. High-skill strategic and creative occupations are likely to see large productivity gains, many middle-skill analytical and administrative jobs face elevated substitution risk, and low-skill roles will experience heterogeneous outcomes (some upskilling, some subject to downward pressure). Rapid deployment, scalable agentic ecosystems, and broad enterprise adoption underpin a faster transition, raising urgent policy needs around reskilling, worker protections, and inclusive governance.

Key Points

• Definition and mechanism
  • Agentic AI can independently coordinate workflows, make multi-step decisions, and produce complex outputs, expanding automation beyond routine tasks to many cognitive roles.
• Distributional effects
  • Winners: high-skill strategic, managerial, and creative roles that can leverage agentic AI for productivity amplification and higher-value outputs.
  • Losers (high risk): many middle-skill analytical, administrative, and coordination occupations that involve predictable multi-step workflows or information-processing tasks.
  • Mixed outcomes: low-skill occupations—some benefit from technology-enabled upskilling and task augmentation; others face competitive pressure and downward wage or employment effects.
• Pace and scale
  • Change is expected faster than previous technological shifts because agentic systems are cheaper to deploy, scale readily across firms, and integrate into existing enterprise ecosystems.
• Policy priorities
  • Immediate focus on reskilling and retraining, strengthened labor protections and transition supports, and inclusive AI governance to mitigate widening inequality.

Data & Methods

• Mixed-method design:
  • Occupation-level task analysis: mapped tasks within occupations to agentic-AI capabilities to identify exposure and complementarity.
  • Economic modeling: simulated substitution vs. amplification effects to project employment and wage impacts, and to estimate the speed of transition under different adoption scenarios.
  • Sectoral case studies: qualitative and quantitative analyses in representative industries to surface heterogeneity in adoption pathways, firm incentives, and organizational impacts.
• Outputs:
  • Identification of occupation groups likely to gain versus those at heightened displacement risk, and scenario-based timing estimates for transitions given deployment cost and adoption-rate assumptions.

Implications for AI Economics

• Research implications
  • Need for refined task-level datasets and models that capture agentic-AI capabilities (planning, tool use, coordination) rather than only routine-task automatability.
  • Importance of sectoral and firm-level heterogeneity in adoption models—aggregate forecasts risk missing concentrated impacts in mid-skill-intensive industries.
  • Dynamic modeling of complementarity (productivity amplification) versus substitution is critical to predict net employment and wage effects.
• Policy and institutional implications
  • Urgency: faster expected transitions increase the importance of near-term policy action.
  • Reskilling strategy: scalable, targeted retraining for mid-skill workers; support for transitions into higher-complementarity roles.
  • Labor protections: portable benefits, unemployment supports, and stronger bargaining/representation options to manage displacement risks.
  • Inclusive governance: regulation and incentives to ensure broad-based access to productivity gains, transparency in enterprise deployment, and mechanisms to monitor distributional outcomes.
• Practical considerations for firms and policymakers
  • Design interventions that foster job-complementary adoption (augmenting human roles) where feasible, while providing transition pathways where substitution is unavoidable.
  • Invest in measurement infrastructures (real-time labor-market monitoring, task-exposure indices) to guide timely policy responses.