Research Article
Redesigning Teaching, Learning, and Assessment in the Age of Generative AI: A Constructive Alignment Perspective for Higher Education
Synthesis: Klapproth (2026) argues that generative AI does not invalidate constructive alignment so much as force an extension of it. The review's claim is that AI use can run down either of two pathways — a scaffold that produces elaboration and reflection, or a shortcut that inflates task performance without building competence — and that which one dominates is decided by instructional design, guardrails and the format through which learning is inferred. The consequence is a specific demand on intended learning outcomes: they must state the conditions of performance, with or without AI, as part of what is being assessed.
Key Findings
- The question has moved from whether to how. The review frames the pedagogical problem as the relationship between intended outcomes, learning activities and the evidence used to judge competence, not as student adoption.
- Two pathways, one design decision. GenAI functions as scaffold or shortcut depending on task design, guardrails and assessment format; the same tool produces both outcomes in the same course.
- Constructive alignment survives but needs a conditional clause. Outcomes have to specify whether the performance is expected with AI, without it, or with it disclosed, because the same task under different conditions measures different things.
- Assessment validity is the pressure point. If the evidence of learning can be produced by the tool, the inference from performance to competence breaks, and the fix is at the format level rather than the detection level.
- Cognitive offloading and feedback research are the mechanisms. The review integrates offloading, scaffolding, self-regulated learning and formative feedback to explain when AI helps and when it substitutes.
The argument
The review is a critical narrative synthesis rather than a systematic one, and its structure follows the alignment triad. On outcomes, it argues for condition specification: an outcome that says students will write an analysis is ambiguous under AI, while one that says they will analyze an unfamiliar case under disclosure and defend the interpretation is not. On activities, it separates tasks where AI use is the object of learning from tasks where AI use would displace the learning, which is a design distinction rather than a prohibition. On assessment, it argues that validity, not authenticity, is the criterion, because an authentic task performed by a tool is still uninformative about the student.
The strongest part of the argument is the claim that the two pathways are not a property of the tool. Elaboration and reflection happen when the task requires the student to explain, compare or defend; inflation happens when the task's evidence can be generated. That is testable, and it locates the instructor's leverage in task design rather than in policy.
What this means for practice
- Instructors: rewrite outcomes to specify the conditions of performance — with AI, without AI, or with disclosed use — so the evidence matches the claim.
- Assessment designers: judge formats by validity under AI conditions. If the evidence can be produced without the learning, the format is the problem.
- Curriculum designers: separate AI-as-object tasks (critique the output) from AI-as-tool tasks (use it to draft), and make the distinction visible in the assignment.
- Instructors: design for elaboration. The scaffold pathway is produced by tasks that require explanation, comparison and defense, not by adding AI to an existing task.
Limitations
A narrative review, not a systematic one, so the evidence base is selected rather than exhaustive and no effect sizes are pooled. The framework is not tested empirically in the paper; it is an argument about how existing theory should be applied, and its claims about which pathway dominates remain hypotheses for design research. The higher-education focus leaves school and professional contexts out. Guidance on disclosure is asserted at the level of principle rather than operational detail.
Citation
Klapproth, F. (2026). Redesigning Teaching, Learning, and Assessment in the Age of Generative AI: A Constructive Alignment Perspective for Higher Education. Education Sciences, 16(1507).