Rethinking LLM Human Simulation: When a Graph is What You Need We identify a large class of simulation problems in which individuals make choices among discrete options, where a graph neural network (GNN) can match or surpass strong LLM baselines despite being three orders of magnitude smaller
Rethinking LLM Human Simulation: When a Graph is What You Need TL;DR: We present a graph-based link-prediction method for discrete-choice human simulation that matches or surpasses LLM-centric human simulation approaches while offering additional advantages
Rethinking LLM Human Simulation: When a Graph is What You Need The paper introduces Graph-basEd Models for Human Simulation (GEMS), a novel framework that rethinks the use of large language models (LLMs) in human simulation tasks by proposing a graph-based alternative
Rethinking LLM Human Simulation: When a Graph is What You Need We introduce Graph-basEd Models for human Simulation (GEMS), which casts discrete choice simulation tasks as a link prediction problem on graphs, leveraging relational knowledge while incorporating language representations only when needed
Rethinking LLM Human Simulation: When a Graph is What You Need We present GEMS to model a large class of LLM-based human simulation problems as link prediction on a graph GEMS learns from the relational structure of choices, and uses a lightweight LLM‑to‑GNN projection only when text is required