Abstract:

In an evolving higher education landscape amid ongoing technological disruption and social uncertainty, surfacing enduring skills that can be applied across disciplines is increasingly recognised as key to helping students navigate their future lives, work and lifelong learning. Skills mapping is among the solutions for surfacing these skills by aligning course learning outcomes (CLOs) with skills. Mapping skills to courses not only empowers students to recognise and articulate their enduring skills, but also provides a common ground for richer, more in-depth conversations about skills. By highlighting strengths and gaps in skill development, skills mapping also supports educators in reflective practice and data-informed curriculum redesign. However, systematically mapping skills across diverse programs remains challenging, as traditional approaches are often manual, thus labour-intensive and inconsistent. To address this issue, we present a novel AI-driven, human-in-the-loop skills-mapping approach developed and verified at an Australian university. This approach first establishes nine enduring skill categories through a systematic review of existing literature and the synthesis of existing government and industry skills taxonomies. Using these categories as a foundation, CLOs are automatically mapped to the nine skills by a machine learning model pre-trained on a dataset of CLO-skill alignment annotated by educators. The model’s output is then verified and refined via human-in-the-loop review by educators of specific courses or programs. Educators’ feedback is iteratively incorporated into the model to continually improve it. Through this AI–human collaboration, the model achieves high mapping accuracy across a broad curriculum. After the piloting phase with over 100 courses in a faculty, over 90% of the automatic mapping results are now validated and approved by educators, thus significantly reducing educators' manual work. After the official deployment, the model achieved over 90% accuracy in mapping skills across more than 1,000 courses within the faculty. Our approach has great potential for large-scale university-wide skills mapping, and is extensible to other curriculum mapping contexts. It can be scaled to align PLOs, CLOs, assessments, and skills, or even annotate courses against external frameworks such as the United Nations Sustainable Development Goals.

Keywords:

Skill, mapping, higher education, course learning outcome.