Limitations and directions for future research

Several limitations should be acknowledged when interpreting these findings. The course was developed and studied within one institutional setting, which shaped available technologies, support structures, and student populations. Other institutions may face different constraints or opportunities. My dual role as instructor and researcher, and my professional background in creative technology and engineering education, also influenced both design choices and interpretations. Reflexive notes and peer feedback were used to examine these influences, but they remain part of the study's context.

This study was scoped as curriculum-design research analyzing instructor-authored artifacts (§2.6). It did not include IRB-approved instruments for measuring student outcomes, and it does not claim findings about student-level learning, motivation, or skill gain. The analysis focused on the curriculum's artifact trail and on course evaluation inputs collected during teaching. It did not track learners over longer periods to see how they applied GenAI in their careers, further studies, or personal projects. In addition, the global offerings involved self-selected participants who chose to enroll in a free GenAI course, which likely shifted the audience composition toward more motivated or curious participants than the general potential learner population.

Future research can build on this work in several directions. Multi-institutional studies with IRB-reviewed protocols could adapt and test the curriculum-design framework in different universities, countries, and disciplines and provide a stronger sense of how generalizable these design principles are. Longitudinal studies, also IRB-reviewed, that follow learners into internships, industry positions, or further education could offer insight into how GenAI skills, ethical considerations, and collaborative habits developed in the course show up in real practice. More focused studies on equity and access, including differences in internet access, device availability, language, disability, and cultural backgrounds, could help refine how modular, learner-centered GenAI curricula might reduce or unintentionally widen participation gaps. Finally, applying a similar curriculum-as-research approach to other rapidly changing domains, such as cybersecurity, climate technology, or robotics, could test whether modular design, continuous feedback, and co-discovery are useful more broadly for teaching disruptive technologies.