A Human-in-the-Loop Framework for AI-Assisted MOOC Production: Managing GenAI Risks in Educational Content Production

Generative Artificial Intelligence (GenAI) is transforming higher edu-cation, shifting the paradigm from simple tool usage to human-AI co-creation of educational content. While this evolution offers potential for developing person-alized and scalable Open Educational Resources (OER), it introduces systemic risks challenging academic integrity, equity, and Open Science principles. This paper problematizes three critical limitations of GenAI—hallucinations, hege-monic bias, and proprietary opacity—and proposes a mitigation framework built on three pillars: technical grounding through Retrieval-Augmented Generation (RAG), critical human oversight via Human-in-the-Loop (HITL), and commit-ment to openness. We present a case study from the University of Florence’s participation in the PNRR-funded ALMA project, where a pilot study (January– July 2025) involving 10 faculty members developed an innovative workflow for educational video production using GenAI tools. The workflow achieved an 83% reduction in production time while maintaining academic rigor through embed-ded expert review. Our analysis demonstrates that a deliberate, human-centered approach can effectively manage GenAI risks, enabling responsible AI adoption within national digitalization initiatives.