Towards scalable curriculum mapping

Comparing human and GenAI alignment of CLOs to professional standards

Abstract

Curriculum mapping is a critical component of accreditation and continuous improvement in education. However, aligning Course Learning Outcomes (CLOs) to professional standards such as Engineers Australia’s Stage 1 Competency Standard remains time-consuming and labour-intensive. This study evaluates the potential of generative AI (GenAI) to support curriculum mapping by comparing automated outputs with expert human judgement. A stratified sample of 141 (10%) first-year CLOs from a national dataset was analysed using both manual review and GenAI (ChatGPT 4o). Each CLO was mapped to the 16 EA Stage 1 Competencies, and outcomes were classified as Match, Manual Only, GenAI Only, or Neither. Overall agreement was 81.2% with particularly strong alignment in professional and personal attributes. Mismatches were most common in technical competencies, where GenAI over-mapped based on keywords or under-mapped due to limited contextual understanding. These findings suggest GenAI can support curriculum review at scale but requires expert oversight for nuanced or discipline-specific outcomes. The work in progress study contributes to the growing literature on AI in education and offers practical insights into hybrid approaches for accreditation and curriculum design.