Year: 2026
Type: conference-paper
Abstract: Personalized game-based learning holds significant educational promise, but creating custom games for individual learners remains prohibitively expensive. This paper presents Play My Way (PMW), a multi-agent system that autonomously generates personalized educational mini-games without human intervention. The architecture coordinates specialized AI components (a conceptualization LLM, a code generation LLM, and a graphics module) through a placeholder-based orchestration mechanism that enables modular collaboration. The system contributes two primary technical advances. First, it demonstrates autonomous multi-agent coordination for educational content generation, where specialized components collaborate to produce complete, playable games from a learner’s hobby and educational topic. Second, novel graphics generation techniques address consistency in game assets: reference-based character generation for visual coherence, skeleton-guided animation for fluid motion across frames, and iterative inpainting for continuous background expansion. A preliminary feasibility study with 15 participants generated 133 playable games at $0.11 per game, confirming technical viability and establishing baseline quality metrics that identify clear directions for improvement.
Institution: Ghent University
Language: English
Subfield: Developmental and Educational Psychology
Field: Psychology
Domain: Social Sciences
SDG: Quality education
Open Access status: green
OpenAlex ID: W7214516435