# General-purpose LLMs as Models of Human Driver Behavior: The Case of Simplified Merging

arXiv: https://arxiv.org/abs/2604.09609 · HTML: https://mlsanity.com/paper/2604.09609/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 18 of its 34 references that are in the map.
- First posted on arXiv 11 Mar 2026.

## Builds on (most cited first, 18 of 18)

- [Language-Guided Traffic Simulation via Scene-Level Diffusion](https://mlsanity.com/paper/2306.06344/) (2023)
- [The Waymo Open Sim Agents Challenge](https://mlsanity.com/paper/2305.12032/) (2023)
- [Receive, Reason, and React: Drive as You Say with Large Language Models in Autonomous Vehicles](https://mlsanity.com/paper/2310.08034/) (2023)
- [Trajeglish: Traffic Modeling as Next-Token Prediction](https://mlsanity.com/paper/2312.04535/) (2023)
- [Building reliable sim driving agents by scaling self-play](https://mlsanity.com/paper/2502.14706/) (2025)
- [Promptable Closed-loop Traffic Simulation](https://mlsanity.com/paper/2409.05863/) (2024)
- [SceneDiffuser++: City-Scale Traffic Simulation via a Generative World Model](https://mlsanity.com/paper/2506.21976/) (2025)
- [WOMD-Reasoning: A Large-Scale Dataset for Interaction Reasoning in Driving](https://mlsanity.com/paper/2407.04281/) (2024)
- [Trajectory Prediction Meets Large Language Models: A Survey](https://mlsanity.com/paper/2506.03408/) (2025)
- [Epistemological Fault Lines Between Human and Artificial Intelligence](https://mlsanity.com/paper/2512.19466/) (2025)
- [GenFollower: Enhancing Car-Following Prediction with Large Language Models](https://mlsanity.com/paper/2407.05611/) (2024)
- [Naturalistic Computational Cognitive Science: Towards generalizable models and theories that capture the full range of natural behavior](https://mlsanity.com/paper/2502.20349/) (2025)
- [TrajTok: Technical Report for 2025 Waymo Open Sim Agents Challenge](https://mlsanity.com/paper/2506.21618/) (2025)
- [Using Models Based on Cognitive Theory to Predict Human Behavior in Traffic: A Case Study](https://mlsanity.com/paper/2305.15187/) (2023)
- [DecompGAIL: Learning Realistic Traffic Behaviors with Decomposed Multi-Agent Generative Adversarial Imitation Learning](https://mlsanity.com/paper/2510.06913/) (2025)
- [The Application of Driver Models in the Safety Assessment of Autonomous Vehicles: A Survey](https://mlsanity.com/paper/2303.14779/) (2023)
- [Modeling Pedestrian Crossing Behavior: A Reinforcement Learning Approach with Sensory Motor Constraints](https://mlsanity.com/paper/2409.14522/) (2024)
- [Active inference as a unified model of collision avoidance behavior in human drivers](https://mlsanity.com/paper/2506.02215/) (2025)

## Abstract (arXiv)

Human behavior models are essential as behavior references and for simulating human agents in virtual safety assessment of automated vehicles (AVs), yet current models face a trade-off between interpretability and flexibility. General-purpose large language models (LLMs) offer a promising alternative: a single model potentially deployable without parameter fitting across diverse scenarios. However, what LLMs can and cannot capture about human driving behavior remains poorly understood. We address this gap by embedding two general-purpose LLMs (OpenAI o3 and Google Gemini 2.5 Pro) as standalone, closed-loop driver agents in a simplified one-dimensional merging scenario and comparing their behavior against human data using quantitative and qualitative analyses. Both models reproduce human-like intermittent operational control and tactical dependencies on spatial cues. However, neither consistently captures the human response to dynamic velocity cues, and safety performance diverges sharply between models. A systematic prompt ablation study reveals that prompt components act as model-specific inductive biases that do not transfer across LLMs. These findings suggest that general-purpose LLMs could potentially serve as standalone, ready-to-use human behavior models in AV evaluation pipelines, but future research is needed to better understand their failure modes and ensure their validity as models of human driving behavior.