Chatterji, A., Csaszar, F. A., Evans, J., Felin, T., Hullman, J., Lakhani, K. R., Sako, M., Zenger, T. (2026). Can AI do strategy? A dialogue and debate. Strategy Science 11(1) 16–30.
On August 22–23, 2025, a group of scholars gathered at the ION Management Science Laboratory at the Sundance Resort in Utah to debate a deceptively simple question: Can AI do strategy? The essays collected here grew out of that two-day conversation, and they don’t arrive at a tidy consensus—which is rather the point. Todd Zenger served as moderator for the dialogue and debate.
The contributors stake out genuinely different positions:
Felipe Csaszar argues yes, AI can do strategy, but the field needs to do its part first: strategy must become more verifiable, with clear benchmarks and performance signals, before AI systems can meaningfully learn and improve at it. The more promising path isn’t a CEO asking a chatbot for advice, but sophisticated multi-agent systems that dramatically expand the search, representation, and aggregation underlying strategic decision-making.
Teppo Felin pushes back, arguing that strategy is fundamentally forward-looking and causal—and that AI, trained on historical data, is structurally incapable of the novel theorizing that real strategic thinking requires.
Jessica Hullman takes a middle path, acknowledging AI’s genuine usefulness for ideation, synthesis, and simulation, while emphasizing that human oversight remains essential precisely because AI outputs are unreliable and strategy is an iterative sensemaking process that humans must own.
Karim Lakhani calls for moving past the philosophical debate altogether and grounding the discussion in empirical evidence—his lab’s field experiments show measurable performance gains from human-AI collaboration, even as they reveal a “jagged frontier” where AI helps on some tasks and actively hurts on others.
Mari Sako highlights that strategy begins with problem framing, a distinctly human activity that tends to get skipped over in discussions of AI-augmented search and aggregation.
James Evans draws a provocative structural parallel between AI systems and organizations, arguing that both function best not through consensus but through internal conflict—and that presenting users with polished AI recommendations may actually degrade strategic judgment by suppressing the productive disagreement that good decisions require.
Ronnie Chatterji closes with a grounded take: AI will absorb parts of what strategists do today, but accountability, persuasion, and organizational context mean it’s more likely to be a complement than a substitute for the foreseeable future.
Taken together, the essays reveal that disagreements about AI’s strategic potential often mask deeper disagreements about what strategy actually is—a computational problem, a causal-theoretic activity, a professional practice, or an organizational process. That conceptual confusion is one important reason the debate matters: how the field answers it will shape not only how strategy scholarship develops, but how organizations actually use AI to make consequential decisions.
Other links: