Understanding the Role of Learning Analytics and Generative Artificial Intelligence on Decision-Making and Learning Design Practice in Higher Education

Synthesis: Instructors use learning analytics and generative AI differently in learning design. Claassen et al. (2026), a quantitative-ethnography study of 11 focus groups with instructors at a large Australian university using Epistemic Network Analysis (ENA) and Self-Determination Theory (SDT) as an interpretive lens, find that LA-informed design discussions center on context, course-level design, and creative Problem Solving, while GenAI discussions center on assessment design and designing for student self-determination. Across both, instructors consistently combine technological affordances with contextual considerations and creative problem-solving. Supporting instructors' basic psychological needs — autonomy, competence, relatedness — fosters creative, effective learning design.

Key Findings

• LA and GenAI support different parts of learning design. In LA discussions, the strongest co-occurrences were between contextual information, course-level design, and creative problem-solving (the technology-role↔context connection was strongest, 0.32). In GenAI discussions, the strongest were between technology role and assessment design (0.26) and technology role and design for student self-determination (0.15).
• Instructors use LA for problem-solving across design granularity. LA (e.g., teaching dashboards) was used in tandem with contextual information to make course-, activity-, and student-support-level adjustments — confirming that LA cannot be meaningfully interpreted without understanding pedagogical intent and context.
• Instructors use GenAI for ideation and assessment development. GenAI was used to create first drafts of assessment elements (case studies, rubrics) and to co-design course elements with students (e.g., co-created assessment rubrics in initial teacher education), guided by their design approach.
• Distinct reasoning patterns. LA discourse emphasized student support in relation to technology more than GenAI did; GenAI discourse elicited stronger emotional responses and centered on assessment design and student self-determination. The technology-role↔student-support connection was notably weaker in GenAI discourse.
• Context and Creativity are central across both. The co-occurrence of technology role, contextual information, and creative problem-solving was high across both LA and GenAI networks — instructors combine tool affordances with context and creativity to solve instructional problems.
• SDT lens: supporting instructor needs matters. Supporting instructors' autonomy, competence, and relatedness fosters creative problem-solving in learning design; collaborative design and shared institutional decision-making support these needs.

What this means for practice

• Instructional designers. Treat Learning Analytics and Generative AI as complementary rather than interchangeable: use LA to diagnose engagement and target support, and GenAI to draft assessment elements and rubrics.
• Instructors. Ground every analytics interpretation in pedagogical intent: in the LA network, technology role co-occurred most strongly with contextual information and creative Problem Solving (0.32), so dashboards cannot be read apart from course context.
• Instructors. Co-design assessment with students using GenAI; the GenAI network linked technology role to assessment design (0.26) and to design for student self-determination (0.15), supporting autonomy and competence.
• Administrators. Support instructors' basic psychological needs — autonomy, competence, and relatedness — through collaborative design, shared decision-making, and transparent information flow, the conditions the SDT lens ties to creative Learning Design.

Limitations

• Participants came from one faculty of one large Australian university — instructors in pre-service teacher education and enabling programs recruited voluntarily — so the findings are not generalizable to other disciplines or institutional settings.
• Fourteen participants contributed to both the LA and the GenAI focus groups, and the 11 groups (21 LA participants, 18 GenAI participants) met online for only 45–60 minutes each.
• The ENA results describe discourse co-occurrence, not observed design behavior; the authors state that connections drawn from low or high co-occurrence are researcher inferences grounded in discourse patterns.
• Coding reliability was initially weak — low pre-moderation K-values for codes such as D.AUTO and I.E., which showed high raw agreement but disproportionately low kappa — and the binary aggregation used in the ENA model does not account for the intensity or repetition of connections within a unit of analysis.

Citation

Claassen, A., Ebbert, D., Kovanović, V., Mirriahi, N., & Dawson, S. (2026). Understanding the Role of Learning Analytics and Generative Artificial Intelligence on Decision-Making and Learning Design Practice in Higher Education. International Journal of Educational Technology in Higher Education, 23, 43.