We Need Strong Preconditions For Using Simulations In Policy
2026-04-09 • Computers and Society
Computers and Society
AI summaryⓘ
The authors explain that simulations using large language models (LLMs) can help policymakers test ideas and predict results, but there are important challenges. These include risks from creating detailed models of people's behavior and difficulties in checking if the simulations are accurate. They suggest three rules to use these simulations responsibly: don't treat models of marginalized groups as just data without considering context, include those groups in the simulation process, and make sure there is clear accountability. They also recommend reporting on how simulations are made and used to build trust and ensure ethical practices.
Large Language ModelsSimulationsPolicymakingAgent-based ModelingValidationEthicsMarginalized PopulationsAccountabilityDual-useHuman Behavior Modeling
Authors
Steven Luo, Saanvi Arora, Carlos Guirado
Abstract
Simulations, and more recently LLM agent simulations, have been adopted as useful tools for policymakers to explore interventions, rehearse potential scenarios, and forecast outcomes. While LLM simulations have enormous potential, two critical challenges remain understudied: the dual-use potential of accurate models of individual or population-level human behavior and the difficulty of validating simulation outputs. In light of these limitations, we must define boundaries for both simulation developers and decision-makers to ensure responsible development and ethical use. We propose and discuss three preconditions for societal-scale LLM agent simulations: 1) do not treat simulations of marginalized populations as neutral technical outputs, 2) do not simulate populations without their participation, and 3) do not simulate without accountability. We believe that these guardrails, combined with our call for simulation development and deployment reports, will help build trust among policymakers while promoting responsible development and use of societal-scale LLM agent simulations for the public benefit.