AI, pedagogy, assessment: Shifting to a design-based pedagogy

Faced with the ease with which students can now generate sophisticated outputs, universities are pressed to confront an uncomfortable question: how can we ensure the integrity and authenticity of student learning when machines can so easily simulate understanding?

My view, informed by collaborative work in 2021 with Najwa Norodien-Fataar, is that the answer lies, not in doubling down on detection and prohibition, but in rethinking the pedagogical architecture of higher education itself.

Our proposal, centred on the notion of e-learning ecologies and design-based pedagogy, based on the 2017 book e-Learning Ecologies: Principles for new learning and assessment, edited by Bill Cope and Mary Kalantzis, offers an opportunity to reposition AI as a partner in a reflexive learning process rather than a threat to be policed.

‘Policy of suspicion’

Current approaches to academic dishonesty are fast becoming obsolete. AI-detection technologies such as Turnitin’s AI-writing indicator are imprecise, frequently generating false positives and negatives.

These tools cannot detect human-copied AI output, paraphrased prompts, or content composed in one interface and repackaged in another. The arms race between detection and evasion is technically unsustainable and educationally impoverished. A policy of suspicion breeds adversarialism; it undermines trust, flattens complexity, and constrains innovation.

Foreground the learning journey

What we require is a shift in the underlying grammar of assessment. In our research, Norodien-Fataar and I have advocated for pedagogical designs that decentre the final product and foreground the learning journey instead.

Design-based pedagogy, as we conceptualise it, positions students, not as passive recipients of information, but as active, agentic knowledge-makers embedded in a dialogic, iterative learning ecology. The digital is not an external add-on; it is entwined with the modalities through which contemporary learners access, process and produce knowledge.

This approach offers powerful resources for mitigating AI-based plagiarism, which some now term ai-giarism. The problem with plagiarism is not only that it violates rules, but that it sidesteps the learning process itself. Design-based assessment disrupts this evasion by embedding students in layered, temporal and socially situated learning experiences.

When learners are required to produce work across stages, such as brainstorming, feedback incorporation, reflective commentary, oral defence, or peer review, there is no single output to outsource. The assessment becomes an activity ecology, requiring cognitive and ethical presence throughout.

For instance, imagine a public health student designing a local campaign on vaccine hesitancy. The assessment begins with a contextual research report, followed by design mock-ups, a peer feedback session, and a final presentation that critically reflects how AI tools were used and evaluated in the process. Here, AI can be part of the toolbox to model audience reception or suggest design frameworks. Still, it cannot complete the task in isolation. The student’s learning process is being assessed, not the product alone.

Plagiarism is mitigated, not by fear of detection, but by making learning impossible to fake because it is embodied, reflective and distributed over time.

In a design-based pedagogical framework, the assessment task is no longer a single product submitted at the end of a module, but a multi-layered developmental process. Learning is shown through a display of artefacts, provision of peer feedback and research notes, AI interaction logs, self-reflections, and oral presentations.

These, together, trace the student’s evolving relationship with the task. Each of these artefacts acts as a window into the learner’s thinking, decisions, and engagement.

Students called on to ‘own’ their thinking

When assessment is structured in this way, dishonesty becomes structurally complex. Students cannot simply copy and paste an AI-generated response and pass it off as their own, because the task requires them to narrate their thinking, explain shifts in understanding, and respond to feedback contingent on their previous submissions.

It is the very distributed nature of the task, its unfolding across time, tools, and modes, that makes it incompatible with singular acts of fabrication. Even if a student were to use AI at one stage, the demand to integrate this use critically, defend it, adapt it in response to new input, and position it within a personal or disciplinary context, reactivates the student’s authorship.

Moreover, this kind of assessment calls upon students to bring themselves into the learning process. They must draw on their experience, positionality and interpretive resources.

A chatbot can simulate argument, but it cannot situate that argument in the student’s own moral reasoning, or account for how it connects to their background, values or emerging worldview. Nor can it respond meaningfully to a lecturer’s probing question in an oral defence or revise an idea in response to class discussion. In these reflexive moments, when students are called to own their thinking, the depth of learning becomes evident.

Rendering plagiarism irrelevant

By making assessment dialogical, temporal, and reflective, we prevent plagiarism and render it pedagogically irrelevant.

The structures of learning, themselves, are the defence. When students are embedded in authentic learning ecologies that require sustained intellectual labour and metacognitive awareness, shortcuts lose their appeal. Integrity becomes, not an external requirement, but an internalised practice.

The integration of reflexive pedagogy is key. As we argued in our 2021 article titled, ‘Towards an approach to pedagogy based on e-learning ecologies in a post-COVID world’, the traditional transmission pedagogy, top-down, linear, and tightly scripted, has failed to cultivate the intellectual independence our current age demands.

Emergency Remote Teaching, which rapidly transferred analogue pedagogies into digital platforms during the pandemic, only exacerbated this failure. Lecturers uploaded content and pre-recorded lectures, but these mirrored pre-COVID-19 epistemologies too often: static, decontextualised, and indifferent to the learner’s interpretive labour. We proposed, instead, a design-based pedagogy that activates students as designers of their own learning pathways, supported through scaffolded activities, formative engagement, and recursive feedback.

When such pedagogies are in place, dishonest or uncritical use of AI becomes detectable and structurally discouraged. A student may be able to use AI to generate a surface-level summary, but they will struggle to explain how it connects to lived experience, to critique its biases, or to defend their intellectual decisions during conversation.

Crucially, this change does not depend on individual lecturers. Institutions must provide the support, guidance, and encouragement for such shifts to occur. This means embedding design-based pedagogy into curriculum review processes, teaching development programmes, and institutional strategy.

Quality assurance frameworks should move beyond their emphasis on syllabus coverage and technical compliance to foreground indicators such as scaffolded learning pathways, multimodal engagement, and student reflection.

Academic staff development

Professional development is essential. Universities should offer training that enables lecturers to reconceptualise assessment beyond the final essay or test.

Interdisciplinary workshops, micro-credential courses, and faculty-based Communities of Practice can provide space for experimentation, co-design and peer mentorship. Lecturers must feel supported intellectually and administratively to redesign their modules in AI-resilient ways, not AI-fearing.

What students should be taught

Equally, students must be educated in AI’s ethics, affordances and limitations. This includes understanding the bias embedded in language models, the superficiality of AI reasoning, and the distinctions between assistance and appropriation. Teaching critical AI literacy becomes a graduate attribute: to know when, how, and why to use these tools in ways that augment one’s intellectual processing.

Institutions should make it a requirement for students to declare AI use in their assignments. Students must explain their AI-based prompts, evaluate the output, and reflect on how it shaped their learning. In doing so, they are being held accountable and, in addition, invited into a deeper metacognitive relationship with knowledge and learning.

There are persuasive examples of this open-ended pedagogical engagement. Students can be tasked with using an AI tool to generate a response to a complex social issue, and then critique its assumptions using course readings and theoretical frameworks.

Another example involves maintaining a design journal that tracks the evolution of a project, including feedback received, AI support used, and design changes made along the way. These practices shift the centre of gravity from what was produced to how it was produced and why, which is the heart of authentic learning.

What is vital is that the institutional narrative changes. Instead of fixating on catching AI use, we must build systems where the value of thoughtful, situated, and reflexive learning is evident and structurally rewarded. Suppose students understand that what matters is how they arrive at their conclusions, not merely what they conclude. In that case, they are far less likely to outsource their work.

Suppose they are engaged, empowered, and invited into the assessment process as co-designers. In that case, they will find meaning that no AI output can substitute.

Rehumanising

This is not a nostalgic return to pre-digital learning, nor a techno-optimist embrace of AI for its own sake. It is a pragmatic and principled approach to rehumanising education in the age of machine intelligence. By embedding assessments in e-learning ecologies, we cultivate epistemic agency, socio-technical awareness, and ethical responsibility.

The future of education belongs to those institutions willing to rethink their assumptions, reinvest in pedagogy, and reimagine the learning journey as something not just protected from plagiarism, but inherently resistant to it. In doing so, we defend our academic integrity.

Aslam Fataar is a research professor in higher education transformation in the department of education policy studies at Stellenbosch University, South Africa.

Commentary articles are the opinion of the author and do not necessarily reflect the views of University World News.