AgentSchool: An LLM-Powered Multi-Agent Simulation for Education

2026-05-28

AI Summary

This study addresses the ethical, institutional, and temporal challenges of validating AI interventions in real educational settings by proposing a large language model–based multi-agent educational simulator. The framework models learning as a state-transition process, integrating student agents endowed with cognitive growth mechanisms, adaptive teacher agents grounded in the Zone of Proximal Development theory, a configurable scenario generator, and a multi-scale simulation engine. Innovatively combining weighted knowledge graphs, a workflow pool for modeling reasoning processes, and explicit misconception mechanisms, the system transcends conventional role-playing simulations. Experimental results demonstrate its capacity to generate diverse trajectories of competence and misconception, and to reproduce classroom social phenomena such as peripheral participation and clique formation, thereby establishing education as a viable long-term, multi-agent coordination testbed.

Abstract

Despite the rapid deployment of LLMs into classrooms, validating educational AI remains uniquely intractable: interventions act on developing learners whose cognitive and social trajectories are irreversibly shaped, while real-world trials are slow, ethically constrained, and institutionally locked. LLM-based educational simulators have emerged as a potential remedy, but many still collapse learning into persona-conditioned role-play and, when optimized only to reproduce existing classrooms, can structurally penalize the institutional novelty that pedagogical reform requires. In this work, we introduce AgentSchool, an LLM-driven multi-agent simulator that models learning as state transition rather than prompted behavior. AgentSchool couples cognitively growable student agents -- equipped with weighted subject knowledge graphs, thinking-workflow pools, and explicit misconceptions -- with adaptive teacher agents that plan, scaffold, and reflect along the Zone of Proximal Development, embedded in a configurable scenery generator that situates instruction within both formal and informal learning fields, and a multi-scale simulator that decouples interaction scale, temporal granularity, and simulation duration. Experiments show that structured student agents produce more differentiated mastery and misconception traces than a baseline simulator, while teacher-agent comparisons show backbone-dependent patterns consistent with ZPD-informed adaptation. Further, AgentSchool generates plausible traces of peripheral participation, clique formation, aggressor-induced cohesion, and opinion-leader emergence consistent with classroom social theories. Beyond its role as an educational research instrument, AgentSchool frames education as a socially meaningful testbed for long-horizon memory, multi-agent coordination, and future institutional reasoning under organizational pressure.

Problem

• educational AI validation
• learning trajectory
• ethical constraints
• institutional barriers
• cognitive development

Innovation

• multi-agent simulation
• cognitive modeling
• Zone of Proximal Development
• knowledge graph
• educational AI