CODE-GEN: A Human-in-the-Loop RAG-Based Agentic AI System for Multiple-Choice Question Generation

📰 ArXiv cs.AI

CODE-GEN is a human-in-the-loop RAG-based agentic AI system for generating multiple-choice questions to develop student code reasoning and comprehension abilities

advanced Published 7 Apr 2026

Action Steps

Employ a human-in-the-loop approach to ensure generated questions are accurate and relevant
Utilize a retrieval-augmented generation (RAG) framework to generate multiple-choice questions
Implement an agentic AI architecture with Generator and Validator agents to produce and assess questions
Integrate course-specific learning objectives to align generated questions with educational goals

Who Needs to Know This

AI engineers and educators on a team can benefit from CODE-GEN as it generates context-aligned multiple-choice questions, and educators can use it to develop student code reasoning and comprehension abilities

Key Insight

💡 CODE-GEN's human-in-the-loop RAG-based approach can effectively generate context-aligned multiple-choice questions for developing student code reasoning and comprehension abilities