Teaching is Not a Download: AI, Memory and Creative Education

As artificial intelligence enters global classrooms with promises of scale, efficiency, and personalisation, creative education finds itself in a precarious position. Unlike STEM disciplines, where problems often appear tractable to automation, fields such as art, design, craft, and cultural studies depend upon pedagogy as situated, embodied, and political labour. This paper examines the epistemic violence that AI systems risk perpetuating in the context of Indian creative education, particularly through acts of forgetting. Drawing on postcolonial theory, memory studies, and critical pedagogy, it argues that we must design AI systems that do not merely simulate teaching but reckon with its historical, cultural, and ethical weight. Without this, the classroom risks becoming another site of digital extraction: clean, quick, and amnesiac.

AI Arrives in the Classroom

Across policy documents, pitch decks, and panel discussions, AI is being heralded as the future of education. “Access for all,” “learning at scale,” “data-driven personalisation” — the language is familiar, and so is its speed. In India, this acceleration is bolstered by the National Education Policy (NEP) 2020, which opens space for AI to enhance teaching–learning experiences, especially in remote, multilingual, and resource-starved contexts.

But behind the promises lies a dangerous assumption: that teaching is a problem to be optimised, and that technology, in particular, AI, can solve it. This assumption not only flattens the meaning of education but also actively threatens the epistemic traditions of creative disciplines. When teaching is reduced to content delivery, learning becomes interface navigation. The teacher becomes a system bottleneck. And memory — especially the slow, unresolved kind — becomes irrelevant.

The urgency of this critique is not technophobic. It is historical. AI systems do not emerge in a vacuum. They are built from data scraped from the past, trained on the biases of the present, and deployed into futures already shaped by inequality. In the context of Indian art and design education, this means that caste, colonialism, and erasure are not outside the machine — they are inside its logic.

Creative Education as Memory Work

In disciplines such as design, architecture, craft, visual culture, and art history, the pedagogical space is not simply one of skill acquisition or content consumption. It is a space where histories are contested, aesthetic canons are questioned, and knowledge circulates in plural, embodied forms.

To teach in this space is to remember what has been forgotten — or forcibly erased. It is to make visible the cracks in the archive. The knowledge of creative practice in South Asia has long been passed down through oral transmission, apprenticeship, and non-textual, sometimes non-verbal, forms. These are forms that resist capture. They do not translate easily into datasets. They are difficult to prompt. And in this resistance lies their value.

The act of teaching, here, becomes one of cultural repair. The teacher is not just a conveyor of syllabus content but a carrier of tensions between modernity and indigeneity, canon and folk, object and process. She performs the archive. She teaches the gap. She models refusal.

If AI enters such a classroom without a theory of memory — or of loss — it will mistake silence for absence, and nuance for error.

The Politics of Forgetting: What AI Cannot Hold

AI systems are, at their core, instruments of memory. But not all memory is created equal. What gets remembered in a dataset depends on what was documented, digitised, and valued. What gets forgotten is everything else.

In colonial and postcolonial educational contexts, forgetting is not an accident. It is structural. The architectural knowledge of Dalit and tribal communities, for instance, was rarely recorded in official blueprints. The devotional paintings of women, the performative rituals of seasonal labour, the ecological wisdom embedded in vernacular design — all existed outside the formal record. In the language of machine learning, these forms are ‘noise’, not ‘signal’.

To build AI systems that ignore this history is to repeat the logic of erasure. The neutrality of the algorithm is a lie. As Ruha Benjamin reminds us, “coded inequity” is not the failure of AI — it is its design principle.[¹] When AI enters the classroom uncritically, it brings with it not just tools, but the weight of past exclusions disguised as objectivity.

For creative disciplines, this flattening is existential. Art history becomes a list of movements. Design becomes a sequence of deliverables. Culture becomes a dataset. The untranslatable, the tactile, the speculative — all get lost in translation.

See content credentials

Pedagogy as Resistance: The Teacher Beyond the Interface

If memory is a political act, then teaching is its most public performance. The classroom is not a neutral space. It is shaped by caste, gender, language, and power. Teachers — particularly in marginalised or interdisciplinary contexts — often carry the burden of holding together fragments of knowledge systems that institutions have systematically devalued.

Their pedagogical work is improvisational, affective, and unfinished. It resists standardisation. A teacher’s pause, a story told at the margins of a lecture, a refusal to categorise — these are not inefficiencies. They are tactics of survival.

AI, in its current form, cannot simulate this. It can replicate patterns, but it cannot inhabit positionality. It can process language, but it cannot hold silence. It can infer, but not remember responsibly.

Here lies the risk: as AI tools become embedded in teaching, the aesthetic of smoothness may overtake the ethics of pedagogy. Teachers may be judged against their AI counterparts — more fallible, less efficient, less consistent. The result is a new kind of extractive economy: one that mines pedagogical labour while discarding the teacher’s role as witness, caretaker, and co-learner.

To respond to this, we need more than “AI for teachers.” We need AI shaped by the histories that teachers carry, especially in fields where the archive itself is contested terrain.

5. Toward a Non-Extractive AI for Education

Designing AI systems for creative education demands a new grammar — one that prioritises care over scale, memory over speed, refusal over completeness.

This means:

• Rejecting the idea of pedagogical neutrality. AI must be trained with positionality in mind, not against it.
• Slowing down system design. Instead of optimising for “engagement,” we must create systems that allow for slowness, confusion, and the unknown.
• Reframing training data as an ethical archive. What is included matters. But so does what is left out — and why.
• Engaging scholars, artists, and educators not just as users, but as co-creators. The system must carry their political commitments, not just their outputs.
• Building in refusal. AI should be able to say: “This cannot be answered. This should not be automated. This must be remembered differently.”

This is not about making AI more human. It is about making AI accountable for the histories it enters.

Refusal is a Form of Pedagogy

We are not at the beginning of the AI-in-education story. We are already inside it. But we still have choices.

We can allow AI to flatten teaching into interaction, to extract culture as content, and to forget with speed. Or we can treat education as a space of historical reckoning — and build systems that remember differently.

This is particularly urgent in India, where the pedagogical labour of holding memory has often fallen to artists, artisans, and cultural practitioners operating on the margins of formal institutions. These are the people AI has historically ignored. They must now shape their future. To teach is not to download. To learn is not to scroll. And to build systems that forget this is to betray the very promise of education.