When Students Know More Than Teachers: AI in Museum Learning

When most institutions talk about bringing AI into education, the instinct is to hand students the tools and watch what they make. The Victoria and Albert Museum took the opposite route. Faced with the question of how to bring AI into its learning programmes, the museum’s Schools and Colleges team made a deliberate decision to start with the teachers.

The resulting three-hour session brought teachers together to explore AI through drawing, collection objects and artist-led experiments.

The reasoning, as David Houston, the team’s leader, explained at the Museum AI Summit, came down to a hard look at reality. AI is already in schools whether anyone feels ready for it or not, and what the team kept hearing from teachers was a deeply mixed picture. Some were using it confidently, folding it into lesson planning and classroom practice. Others didn’t have the time or the confidence to engage with it at all. Beneath that sat the problem of access: some students arrive with powerful devices and tools at home, others with almost nothing. Rather than try to recreate that gap inside the museum, the team chose to support the people who have to make sense of all this for a room full of children at a time.

Why the V&A started with teachers

There’s a tidy story that students are ahead of their teachers on AI, and in a narrow sense it holds. Plenty of students know which tools to use, how to prompt them, and how to get a result quickly. But as Houston put it, being able to generate something is not the same as understanding it. Much of that student use is also happening with no guardrails at all.

Teachers, meanwhile, are being asked to respond at speed. Schools are writing AI policies while the staff expected to follow them are still working out what AI even is. The gap, Houston argued, isn’t really about knowledge. It’s about confidence, understanding, and how to approach the subject in the classroom at all.

That was too tall an order for a learning team to take on alone, and the museum was candid that it wasn’t the expert here. So it built the work around two people who are.

A residency that treats AI as a material, not an oracle

This is the third year of the V&A’s Adobe Creative Residency, supported by the Adobe Foundation as part of a five-year partnership. Each year the museum embeds three artists across its learning programmes, where they develop their own practice while working directly with schools, families and young people, using the collections as a starting point. This year the artist Maria Than joined the Schools and Colleges team, with a residency focused on AI and digital worlds.

Than is a multidisciplinary artist, designer and educator, and her route into AI is a world away from the breathless version usually on offer. She works with creative AI alongside augmented reality, physical computing, moving image, illustration and digital design, often to tell stories about Vietnamese and Tibetan Buddhism and the experiences of people who have migrated away from their homelands. As part of the residency she is also chasing a genuine mystery: the origins of the “Pure Land” digital images that hang in so many Asian families’ homes, temples and restaurants, pictures instantly recognisable to a generation of Southeast Asian immigrants, yet made by artists no one can name.

What matters for the classroom is the philosophy she brings. Than is building an AI manifesto, a set of values and rules to create by, shaped by Buddhist principles: keep your working transparent, and put care and intention ahead of the final output. She is drawn to the idea of “Slow AI,” a counterweight to the Silicon Valley reflex of speed, scale and profit, in the spirit of slow food and slow fashion. It is an approach that looks back to heritage, history and even past failures, and that deliberately makes room for perspectives we don’t hear from often enough, echoing the sociologist Ruha Benjamin’s call to imagine the worlds we cannot live without even as we dismantle the ones we cannot live within.

Crucially, Than does not treat AI as a virtual assistant you hand your thinking to. She treats it as an artist’s material or tool, something reached for with intention, after the looking, the thinking and the drawing are already done. Sketching, she says, is always her starting point: it is how she builds the skeleton of an idea before any tool gets near it.

Building the Teachers’ Sanctuary

To turn this into something useful, the museum paired Than’s practice with classroom reality, bringing in Trudi Barrow, a former Design & Technology head of department who was an early adopter of AI in her own teaching and has worked to bring it into the classroom responsibly. Than’s edge as a practitioner plus Barrow’s hard-won classroom experience gave teachers a voice they could trust.

When the doors opened for a three-hour session, the lack of clarity was striking. Teachers were being asked to think about AI in terms of plagiarism and acceptable use, often without the underlying knowledge to do so confidently. The Department for Education has produced detailed guidance, Houston noted, but most teachers have neither the budget nor the time to hunt for relevant professional development. So the session used the museum’s collection as a springboard for AI that teachers could actually use back in school.

The concerns came thick and fast, and they were real: the impact on creativity, the flattening effect of AI “slop,” a genuine fear that human judgement might disappear from how students are assessed as they enter the workplace, and the environmental cost of the technology.

Drawing the AI black box

The session didn’t begin with a tool. It began with a pen.

Teachers were asked to draw what they thought happens inside the “AI black box,” the unknown space where the software does whatever it does. The exercise was emotional rather than technical. Some drawings were anxious: chaos, lost ownership, in one case a comic-book robot making off with a swag bag of data. Others were gentler, with clouds and arrows linking patterns together. A few were simply whimsical, puppies and all, an honest admission that people often use very powerful software in very playful ways.

From there the team could unpack what AI actually is: pattern recognition learned from training data, which can include human-labelled examples without depending on them in every model. They walked teachers through the idea of latent space, the way AI turns images and ideas into sets of numbers and places similar things near each other, then navigates those relationships to generate something new in response to a prompt. Demystifying the box, Houston said, took much of the dread out of the room.

From sculpture sketch to AI experiment

Then came the obvious question: how do a traditional museum and its very traditional objects fit into any of this? The team answered it with about the most traditional objects it has, the statues in the Sculpture Gallery.

Teachers sketched a sculpture that inspired them by hand, and those sketches became the seed for image-to-image tools such as NewArc and TouchDesigner to iterate on. This is where Barrow introduced an idea worth holding onto: creative fidelity. The principle is that AI should stay true to the student’s original vision. It responds to ideas; it does not overwrite them. The human idea stays intact, and the tool serves it rather than the reverse.

A related approach appears in Kunsthalle Praha’s student AI workshops, which keep exhibition objects at the centre of students’ experiments.

The collection gave the session a shared starting point. Before teachers experimented with image generation, they had already chosen an object, looked closely and made a drawing of their own. That sequence made creative fidelity something they could discuss through a specific piece of work: what had the tool retained, and what had it changed? Alongside the black-box exercise, it connected two different questions teachers were bringing into the room: how AI works and how to keep a learner’s intention visible when using it.

What teachers took from the three-hour session

The shift over the session was the part that clearly moved both speakers. At the start there was hesitation, even dread, about what AI might do to creativity and to a student’s sense of creative identity. By the end there was optimism, a flurry of ideas and a fair amount of laughter, with teachers swapping plans for Monday morning, from generating characters for storytelling sessions to building starting points their students could push off from.

Than felt it too. At the live AI station, she said, teachers were relentless in the best possible way, testing materials, pushing ideas further, feeding their drawings in to see what came back. She recognised something of herself in it, that giddy, almost evangelical thirst for knowledge you get when a new way of working opens up. What made the difference, she suspects, was showing different ways of working with the tools and keeping drawing and making in the mix throughout, so the technology felt like one option among many rather than the only show in town.

What teachers actually need

The lesson the team kept relearning was that teachers don’t need more toolkits. They need time, space and frameworks to work out how this applies in their own classroom. They are engaged; they just need support to turn thinking into practice.

There was real promise, too, for students with special educational needs. AI can build confidence in creativity and prototyping, Houston suggested, offering a route to professional-looking outputs that take some of the anxiety out of whether an idea is good enough.

Than was careful to hold an honest counterweight alongside the excitement. AI isn’t neutral. These tools are made by people, shaped by human decisions and a great deal of often-invisible labour, a point driven home by a booklet in the V&A’s collection, Anatomy of an AI System by Kate Crawford, Vladan Joler and their collaborators, which lays bare the human work, energy and resources behind a single smart speaker. Understanding that, she argued, is part of understanding the tools.

Protected time for experimentation also runs through Cincinnati Museum Center’s staff learning community.

Why museums still matter here

When students know more than their teachers, Houston suggested, the real issue isn’t that students are ahead. It’s what we mean by “knowing.” Being able to generate something is not the same as understanding it, or knowing where it should and shouldn’t be used. That, he argued, is exactly where teachers and museums still matter: not in keeping up with the pace, but in helping everyone slow down enough to make sense of these tools.

The museum has had to wrestle with the same questions on a larger scale, including how to use AI safely with its audiences when the output can’t always be predicted, and how to make an AI offering genuinely particular to this museum and this collection. Its answers so far include inviting AI-assisted entries into the national schools competition V&A Innovate, building teacher resources, and running an online talk where students can see Than’s practice in action.

What has stayed constant, both speakers stressed, is the centre of gravity: creativity, intention and critical thinking. Not promises about what AI will do for us, but the slower, more honest work of figuring out how to use it well.

–

This article is based on a presentation given by David Houston (Team Leader, Schools and Colleges, V&A) and Maria Than (Adobe Artist in Residence, V&A) at the Museum AI Summit on May 27 2026.