Evolution in Simulation: AI-Agent School with Dual Memory for High-Fidelity Educational Dynamics

TL;DR

The paper addresses the challenge of realistically simulating complex educational dynamics with multiple interacting participants. It introduces the AI-Agent School (AAS), a multi-agent platform that uses a Zero-Exp self-evolving mechanism and a dual memory base (Experience and Knowledge), each with Short-term and Long-term components, to drive autonomous improvement via an experience-reflection-optimization loop. An extensive environment, role design, and memory-action taxonomy underpin the simulation, and nine memory configurations are evaluated with automated ROUGE-L metrics and expert human judgments. Results show that the full memory configuration yields the highest alignment with ground-truth educational interactions, demonstrating the viability of high-fidelity educational digital twins and data-driven educational insights for policy and practice.

Abstract

Large language models (LLMs) based Agents are increasingly pivotal in simulating and understanding complex human systems and interactions. We propose the AI-Agent School (AAS) system, built around a self-evolving mechanism that leverages agents for simulating complex educational dynamics. Addressing the fragmented issues in teaching process modeling and the limitations of agents performance in simulating diverse educational participants, AAS constructs the Zero-Exp strategy, employs a continuous "experience-reflection-optimization" cycle, grounded in a dual memory base comprising experience and knowledge bases and incorporating short-term and long-term memory components. Through this mechanism, agents autonomously evolve via situated interactions within diverse simulated school scenarios. This evolution enables agents to more accurately model the nuanced, multi-faceted teacher-student engagements and underlying learning processes found in physical schools. Experiment confirms that AAS can effectively simulate intricate educational dynamics and is effective in fostering advanced agent cognitive abilities, providing a foundational stepping stone from the "Era of Experience" to the "Era of Simulation" by generating high-fidelity behavioral and interaction data.

Figures (7)

• Figure 1: Structural diagram of AAS.
• Figure 2: Zero-Exp mechanism
• Figure 3: Data building process
• Figure 4: GPT-4o Automated Evaluation Result
• Figure 5: Qwen3-235B-A22B Automated Evaluation Result