Stop Drawing Scientific Claims from LLM Social Simulations Without Robustness Audits

• URL: http://arxiv.org/abs/2605.18890v1
• Date: Sun, 17 May 2026 00:21:53 GMT
• Title: Stop Drawing Scientific Claims from LLM Social Simulations Without Robustness Audits
• Abstract summary: Generative agents bring new expressive power to agent-based modeling.<n>Small perturbations that appear minor to researchers can cascade into macro-level outcomes.<n>We introduce TRAILS, a robustness-audit taxonomy spanning three levels of simulation design.
• Score: 11.902271500463934
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: The scientific claims drawn from LLM social simulations should be no stronger than the robustness audits that support them. Generative agents bring new expressive power to agent-based modeling, enabling simulations of collective social processes like cooperation, polarization, and norm formation. Yet they also introduce complexity through additional architectural choices, such as agent specification, memory representation, interaction protocols, and environment design. Small perturbations that appear minor to researchers can cascade into macro-level outcomes through repeated interaction, creating a "butterfly effect." Consequently, scientific claims drawn from LLM social simulations may reflect implementation artifacts rather than the social mechanisms being modeled. We support this position with two case studies: a repeated Prisoner's Dilemma and a social media echo chamber simulation. Across multiple models, minor perturbations in persona format and game-instruction framing shift cooperation rates by up to 76 percentage points, while network homophily and hub assignment produce significant and consistent shifts in polarization metrics. We also find that sensitivity is unevenly distributed across both architectural choices and model families: the same perturbation that produces the 76 pp shift in one frontier model only shifts another by 1 pp. Robustness is therefore a property that should be measured per claim and per model, not assumed. To address this validation gap, we introduce TRAILS (Taxonomy for Robustness Audits In LLM Simulations), a robustness-audit taxonomy spanning three levels of simulation design: agent (micro-level), interaction (meso-level), and system (macro-level). We call for robustness to become a first-order validation requirement before LLM social simulations are used to explain mechanisms, evaluate interventions, or inform decisions.

Related papers

• Will Scaling Improve Social Simulation with LLMs? [87.15677770205751]
  Large Language Model (LLM) social simulations are a promising research method, but they are not yet faithful enough to be adopted widely.<n>We use scaling laws to study the relationship between LLMs' compute scale, general capability benchmarks, and the fidelity of social simulation.
  arXiv  Detail & Related papers  (2026-07-02T17:30:38Z)
• Simulate, Reason, Decide: Scientific Reasoning with LLMs for Simulation-Driven Decision Making [50.05403024674141]
  MechSim is a mechanism-grounded neuro-symbolic reasoning framework for executable scientific simulators.<n>Our framework represents simulators through a shared structured schema capturing assumptions, variables, mechanism dependencies, and execution traces.<n>We evaluate our approach across multiple high-stakes domains and show that it improves mechanism-level explanation quality, simulator analysis, and downstream decision-making reliability.
  arXiv  Detail & Related papers  (2026-06-03T06:36:51Z)
• SYN-DIGITS: A Synthetic Control Framework for Calibrated Digital Twin Simulation [9.671885065974857]
  SYN-DIGITS is a principled and lightweight calibration framework for AI-based persona simulation.<n>It learns latent structure from digital-twin responses and transfers it to align predictions with human ground truth.<n>It supports both individual-level and distributional simulation for previously unseen questions.
  arXiv  Detail & Related papers  (2026-04-08T18:50:01Z)
• Position: AI Agents Are Not (Yet) a Panacea for Social Simulation [62.891898598784415]
  Recent advances in large language models (LLMs) have spurred growing interest in using LLM-integrated agents for social simulation.<n>This position paper argues that LLM-based agents are not (yet) a panacea for social simulation.
  arXiv  Detail & Related papers  (2026-02-19T06:35:07Z)
• PhysicsAgentABM: Physics-Guided Generative Agent-Based Modeling [47.029742241618635]
  Large language model (LLM)-based multi-agent systems enable expressive agent reasoning but are expensive to scale and poorly calibrated for timestep-aligned state-transition simulation.<n>We propose PhysicsAgentABM, which shifts inference to behaviorally coherent agent clusters.<n> Experiments across public health, finance, and social sciences show consistent gains in event-time accuracy and calibration over mechanistic, neural, and LLM baselines.
  arXiv  Detail & Related papers  (2026-02-05T18:59:01Z)
• SimBench: Benchmarking the Ability of Large Language Models to Simulate Human Behaviors [58.87134689752605]
  We introduce SimBench, the first large-scale, standardized benchmark for a robust, reproducible science of LLM simulation.<n>We show that even the best LLMs today have limited simulation ability (score: 40.80/100), performance scales log-linearly with model size.<n>We demonstrate that simulation ability correlates most strongly with deep, knowledge-intensive reasoning.
  arXiv  Detail & Related papers  (2025-10-20T13:14:38Z)
• Integrating LLM in Agent-Based Social Simulation: Opportunities and Challenges [0.7739037410679168]
  The paper reviews recent findings on the ability of Large Language Models to replicate key aspects of human cognition.<n>The second part surveys emerging applications of LLMs in multi-agent simulation frameworks.<n>The paper concludes by advocating for hybrid approaches that integrate LLMs into traditional agent-based modeling platforms.
  arXiv  Detail & Related papers  (2025-07-25T15:15:35Z)
• LLM-Based Social Simulations Require a Boundary [3.351170542925928]
  This position paper argues that large language model (LLM)-based social simulations should establish clear boundaries.<n>We examine three key boundary problems: alignment (simulated behaviors matching real-world patterns), consistency (maintaining coherent agent behavior over time), and robustness.
  arXiv  Detail & Related papers  (2025-06-24T17:14:47Z)
• Modeling Earth-Scale Human-Like Societies with One Billion Agents [54.465233996410156]
  Light Society is an agent-based simulation framework.<n>It formalizes social processes as structured transitions of agent and environment states.<n>It supports efficient simulation of societies with over one billion agents.
  arXiv  Detail & Related papers  (2025-06-07T09:14:12Z)
• Sense and Sensitivity: Evaluating the simulation of social dynamics via Large Language Models [27.313165173789233]
  Large language models have been proposed as a powerful replacement for classical agent-based models (ABMs) to simulate social dynamics.<n>However, due to the black box nature of LLMs, it is unclear whether LLM agents actually execute the intended semantics.<n>We show that while it is possible to engineer prompts that approximate the intended dynamics, the quality of these simulations is highly sensitive to the particular choice of prompts.
  arXiv  Detail & Related papers  (2024-12-06T14:50:01Z)
• GenSim: A General Social Simulation Platform with Large Language Model based Agents [111.00666003559324]
  We propose a novel large language model (LLMs)-based simulation platform called textitGenSim.<n>Our platform supports one hundred thousand agents to better simulate large-scale populations in real-world contexts.<n>To our knowledge, GenSim represents an initial step toward a general, large-scale, and correctable social simulation platform.
  arXiv  Detail & Related papers  (2024-10-06T05:02:23Z)
• Investigating the Robustness of Counterfactual Learning to Rank Models: A Reproducibility Study [71.04084063541777]
  Counterfactual learning to rank has attracted extensive attention in the IR community.<n>Models can be theoretically unbiased when the user behavior assumption is correct and the propensity estimation is accurate.<n>Their effectiveness is usually empirically evaluated via simulation-based experiments due to a lack of widely available, large-scale, real click logs.
  arXiv  Detail & Related papers  (2024-04-04T10:54:38Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.