Designing Effective MOOCs Through a Custom GPT Framework: A Case Study on Human-AI Collaboration in Instructional Design at the University of Turin

In the context of increasing demand for scalable, inclusive, and pedagogically coherent digital education, this paper presents MOOCLab: a customized version of ChatGPT developed at the University of Turin to support the design of MOOCs (Massive Open Online Courses) on Moodle. Created through a collaboration between the university’s E-learning staff and ICT units, MOOCLab was tailored to function as an instructional design assistant aligned with key pedagogical models—including the ADDIE: Analyze, Design, Develop, Implement. Evaluate framework, Bloom’s Taxonomy, Universal Design for Learning (UDL), Dublin Descriptors, ESCO (European Skills, Competences, Qualifications and Occupations) standards and WACG (Web Content Accessibility Guidelines). Current MOOC tools and generic AI assistants lack pedagogical alignment and accessibility, resulting in low completion rates and limited scalability. MOOCLab was designed to guide instructors, both novice and experienced, through all phases of MOOC design by providing adaptive, domain-sensitive support. From defining learning objectives to selecting Moodle and H5P activities, to producing accessible multimedia content, MOOCLab offers iterative, evidence-based, and context-aware recommendations. This paper details the assistant’s technical architecture, operational logic, and customization process, and discusses its potential as a replicable model for AI-powered instructional design in higher education.