Generative AI can both sharpen and dull corporate strategic foresight depending on how it’s deployed: four distinct collaboration roles—from embedded Ally to disruptive Provoking Catalyst—systematically alter attention, search scope and cognitive friction, so sequencing and governance determine whether firms gain broader, more robust scenarios or fall prey to narrowed frames and overconfidence.

Main Finding

GenAI can reshape organizational strategic foresight in four distinct, ideal-typical roles—Resilience Builder, Ally Insider, Provoking Catalyst, and Critical Outsider—each altering attention, search breadth, and cognitive friction in ways that systematically affect scenario diversity, weak-signal detection, assumption surfacing, and the quality of strategic recommendations. These roles are contingent collaboration modes that organizations can sequence across foresight stages; depending on design and timing, GenAI can either strengthen or weaken foresight outcomes.

Key Points

• Microfoundations: integrates bounded rationality, dual-process theory, and sensemaking to explain how GenAI changes cognitive constraints and interaction dynamics in foresight tasks.
• Mechanisms:
  • Attention: GenAI can focus or diffuse human attention (e.g., by surfacing candidate issues or by filtering inputs).
  • Search breadth: GenAI can expand the range of evidence and scenarios explored (via large-data pattern recognition) or narrow search via prompt/algorithmic framing.
  • Cognitive friction: GenAI can reduce friction (speeding convergence) or increase it (provoking challenge and reflection).
• Four ideal-typical roles (summarized effects):
  • Resilience Builder: Integrates with existing workflows to stabilize and broaden scenario coverage; increases robustness and coverage but may reinforce dominant frames.
  • Ally Insider: Deeply embedded assistant that speeds analysis and operationalizes human heuristics; high efficiency and alignment but risk of reinforcing existing biases and reducing independent challenge.
  • Provoking Catalyst: Deliberately generates divergent, disruptive perspectives; raises scenario diversity and weak-signal detection but can increase noise and require human filtering.
  • Critical Outsider: Acts as a disconfirming analyst, surfacing hidden assumptions and counterfactuals; increases cognitive friction and assumption surfacing but may slow consensus formation.
• Sequencing: Roles are not mutually exclusive—best practice is contingent sequencing across foresight phases (e.g., use Provoking Catalyst or Critical Outsider early to surface alternatives and assumptions, then Ally Insider/Resilience Builder to refine and operationalize).
• Conditionality: Effects depend on task design, prompt framing, human expertise level, organizational incentives, data inputs, and governance structures; misalignment can produce overconfidence, narrow search, or deskilling.

Data & Methods

• Methodological approach: conceptual/theoretical study combining a targeted literature review with microfoundational theory integration (bounded rationality, dual-process, sensemaking).
• Scenario method: applied a deductive 2 × 2 matrix (theory-driven axes derived from how GenAI affects attention/search breadth and cognitive friction) to generate four ideal-typical collaboration modes.
• Evidence type: theoretical synthesis and deductive scenario construction; no primary empirical or experimental data reported. The framework is presented as a basis for future empirical validation.

Implications for AI Economics

• Task complementarities and substitution: Different GenAI roles imply varying degrees of complementarity with human foresight skills. Embedded roles (Ally Insider, Resilience Builder) tend to substitute routine cognitive effort but complement higher-order judgment; critical/creative roles (Provoking Catalyst, Critical Outsider) complement human deliberation and judgment more directly.
• Returns to expertise and skill-biased change: Organizations that sequence GenAI to preserve cognitive friction and surfacing of assumptions may sustain returns to high-level strategic expertise; poor design risks deskilling and compressing wage premia for foresight specialists.
• Organizational capital and investment: Firms must invest in workflow design, prompts, data governance and training to capture GenAI's potential—investment returns will hinge on role choice and sequencing across foresight stages.
• Productivity vs. risk trade-offs: GenAI can raise foresight productivity (faster scenario generation, broader search) but also create economic risks via overconfidence, weaker diversity of views, or systemic amplification of blind spots—affecting firm-level risk-taking and market dynamics.
• Measurement and valuation: Evaluating GenAI’s economic value requires new metrics (scenario diversity, weak-signal detection rates, assumption surfacing frequency, downstream decision quality), not just throughput or cost savings.
• Policy and regulation: Regulators and governance frameworks should consider how role design affects systemic risks (e.g., synchronized strategic blind spots across firms) and ensure transparency, auditability, and incentives that preserve deliberative friction where needed.
• Research agenda: Empirical tests of the four roles, measurement of sequencing effects on firm performance, labor market impacts of changing foresight tasks, and experiments on prompt/workflow design to optimize complementarities and mitigate risks.