Doctoral Research Project Summary: The Maher AI Agent

Core Project Concept

The Maher AI Agent is an innovative doctoral research project designed to shift Artificial Intelligence in Education (AIEd) away from answers-on-demand tools that encourage cognitive laziness. Instead, it introduces a Socratic conversational AI agent that acts as an individualized thinking coach for 12-to-13-year-old students (6th grade) learning mathematics. By refusing to give direct solutions and dynamically adjusting its guidance based on student mistakes, Maher fosters critical and creative thinking during a critical developmental phase of mathematical abstraction.

Key Innovations

• Deterministic Socratic Engine (DAG Motor): To eliminate the risk of mathematical hallucinations common in standard LLMs, Maher uses a Hybrid Directed Acyclic Graph architecture. This ensures rigorous, three-level Socratic scaffolding (Focusing, Decomposing, Metacognitive Anchoring).
• Multimodal Textbook Ingestion: Using a Visual-to-Code pipeline, the agent can scan, analyze, and adapt to the curriculum of any standard school textbook in real time.
• Field-Ready Classroom Implementation: The system is built for real-world school constraints, featuring a rapid rotation protocol via QR/NFC badges for shared-device classrooms, a fully offline mode with AES-256 local encryption for data privacy (GDPR compliance), and automated xAPI data tracking to precisely measure student learning gains and the gradual reduction of AI assistance (fading index).