Citation: Csaszar, F. A. (2025). Unbounding rationality: Why AI is a fundamental issue for strategy. Available at SSRN: https://ssrn.com/abstract=5454634.
Cite
Paper highlights
Much of strategy theory begins with bounded rationality: people cannot examine every alternative, hold a complete model of the environment, or combine unlimited information. AI relaxes those bounds by changing the feasible amount of search, the complexity of representations, and the scale at which judgments can be aggregated.
The chapter calls this unbounding rationality. The term does not imply perfect rationality or the end of human judgment. It describes a moving boundary: cognitive work that was prohibitively expensive becomes cheap enough to perform routinely. The relevant strategist is therefore a human–AI system, with people orchestrating and auditing analysis while machines expand what the system can search, represent, and aggregate.
How AI changes the strategy process
- Search: LLMs and agentic systems can generate, compare, and criticize far more alternatives than a management team could examine directly.
- Representation: AI can build and update richer models from text, data, scenarios, and causal claims instead of forcing every problem into a small static framework.
- Aggregation: organizations can combine human judgments with multiple AI roles, simulated stakeholders, or creator–critic systems without reproducing every social constraint of a committee.
The immediate gain is not an all-knowing machine strategist. It is a decision system that performs more cognitive work before leaders commit. Human attention moves toward specifying problems, choosing objectives, checking evidence, resolving value conflicts, and owning consequences.
Two objections
- If every firm has AI, it cannot create advantage. LLM access may become common, but proprietary data, AI-native processes, complementary assets, and adaptation speed remain uneven.
- AI is backward-looking and cannot create novel strategy. Historical training data constrain models, but novelty can also emerge from search, recombination, and simulation. Whether it produces useful foresight is an empirical question.
Research agenda
The chapter calls for prospective forecasts, shared benchmarks, simulations, and other artifacts that make strategic performance testable. If strategy scholars define good strategizing precisely enough to evaluate, they can help shape the systems managers will use.
How to use this chapter
Cite this for
- The concept of unbounding rationality in strategy.
- The argument that AI relaxes constraints on strategic search, representation, and aggregation.
- A view of strategy work as a human–AI decision system rather than a standalone human or machine decision-maker.
Useful for teaching
- Bounded rationality as the theoretical starting point for why AI matters in strategy.
- Why expanding cognitive capacity is not the same as perfect rationality or executive replacement.
- How prospective benchmarks and simulations could make strategic performance more testable.
Careful claim
Unbounding rationality means that some cognitive work that was previously too expensive becomes feasible; it does not imply perfect rationality, automatic competitive advantage, or the removal of accountable human judgment.
Abstract
Recent breakthroughs in artificial intelligence represent a fundamental discontinuity for the field of strategy. This chapter argues that AI strikes at the heart of strategy’s foundational assumption: bounded rationality. By offering a path toward “unbounding rationality”—systematically relaxing the cognitive constraints that have shaped organizational decision-making since Simon—AI promises to transform both strategy process (how strategic decisions are made) and strategy content (the sources of competitive advantage). The chapter builds this case by presenting affirmative arguments grounded in AI’s demonstrated mastery of complex cognitive tasks, from creative problem-solving to sophisticated language understanding.
It then directly confronts two powerful objections: that AI’s widespread availability will homogenize rather than differentiate firms, and that AI lacks the forward-looking creativity essential for strategic thought. Finally, it proposes a concrete research agenda centered on computational strategy processes, predictive foresight, verifiable benchmarks, and new forms of scholarly contribution. Embracing this program can sharpen theory, guide practice, and position the field to lead in an era of radical, ubiquitous intelligence.
Other links
Last updated 2026-06-21