Published open access in the Strategic Management Journal.
Read the article · Download the publisher PDF
Large Language Models (LLMs) offer strategy researchers a new way to run simulated experiments with multiple agents and strategic interdependencies. We introduce a framework for using LLM-based AI agents as synthetic subjects for rapid, low-cost prototyping of human experiments and for generating hypotheses.
We apply the framework to the exploration-exploitation dilemma. The AI-agent simulations reproduce patterns observed in human experiments; varying the setup then reveals where those patterns weaken or reverse. This makes the framework useful for iterating on research designs, clarifying boundary conditions, and surfacing new hypotheses for subsequent validation.
Publication
Matteo Tranchero, Cecil-Francis Brenninkmeijer, Arul Murugan, and Abhishek Nagaraj. 2026. “Simulating strategic interactions with AI agents.” Strategic Management Journal. https://doi.org/10.1002/smj.70112
An earlier version, Theorizing with Large Language Models, circulated as NBER Working Paper No. 33033.
BibTeX
@article{tranchero2026simulating,
title = {Simulating strategic interactions with AI agents},
author = {Tranchero, Matteo and Brenninkmeijer, Cecil-Francis and Murugan, Arul and Nagaraj, Abhishek},
journal = {Strategic Management Journal},
year = {2026},
doi = {10.1002/smj.70112},
url = {https://doi.org/10.1002/smj.70112}
}