This is a draft in development. It is being tested with the sector, and the wording will change. Endorse the approach and you will be asked to help rewrite it.
Learning from Objects in the Age of AI
The proposition is in development. Endorsing means helping to shape it.
In an age of artificial fluency, children need opportunities to practise judgement. In a culture of infinite content, they need to know how claims relate to evidence. In a system increasingly able to produce answers on demand, they need slow looking, shared talk, sensory engagement, uncertainty and real questions.
The Real Thing is not an argument for objects over technology. It is an argument for evidential enquiry and human judgement in a world of increasingly synthetic representation.
Thirty-five years after English Heritage published A Teacher's Guide to Learning from Objects, and more than three decades after Eilean Hooper-Greenhill's The Educational Role of the Museum, the case for learning from objects has not faded. It has sharpened. But the challenge is no longer simply to argue that real objects matter. Authenticity alone does not produce learning.
“The case for learning from objects has not faded. It has sharpened.”
A story told to a child may be memorable. A question pursued through an object can be transformative.
If the question has already been answered for the learner, where is the learning?
A replica that can be touched and dismantled may provoke richer reasoning than an original behind glass. A digital reconstruction may reveal what the object cannot. An AI-generated image, used critically, can itself become material for serious discussion about evidence and trust.
So the important distinction is not digital versus real. It is consumption versus judgement.
What material evidence offers is something more precise: an opportunity to observe, question, infer, test claims, encounter uncertainty and revise judgement against evidence that places constraints on the stories we tell about it. A real fragment can tell you less than a convincing reconstruction. But what it tells you can be tested, questioned and built upon.
“The Real Thing is not a campaign for real objects. It is a campaign for the kind of thinking we can teach through material evidence.”
This is not an argument against storytelling. Story is powerful in museum learning. It creates connection, imagination, memory and a reason to care. But it is worth testing where we think the educational value actually lives. If storytelling about objects were the point of museum education, generative AI would already have solved museum education. It can generate fluent, plausible, emotionally engaging narrative about almost anything, instantly. Nobody thinks it has.
Story opens the door. The object is the evidence on the other side.
AI can support this work. It can widen access, enable comparison, reconstruct, generate material worth interrogating and prompt better questions. The boundary is simple. If AI performs the judgement for the learner, the learner is no longer practising judgement.
There is excellent practice across this sector. There is contemporary scholarship. There are practitioners who know exactly how to do this. What is missing is a reliable sector-wide route through which this pedagogy is articulated, taught at entry, honed in practice and refreshed across a career. At the exact moment when evidence, representation, provenance and authenticity have become harder educational concepts, much of the professional knowledge we need remains tacit and unevenly transmitted. So this is not an attempt to invent a new pedagogy. It is an attempt to rearticulate and renew an inherited one for new educational conditions.
The manifesto has four parts, and they answer four different questions. Principles are what we believe. Commitments are what signing asks us to do. Refusals are the lines we will not cross. And outputs are what the sector actually needs next. The first three are drafted. The fourth is deliberately open. All of it is in draft. It goes to the sector for the first time at the GEM Conference 2026, where practitioners will be asked to rank the principles, discard one, audit the commitments against their own practice and mark the refusals they hesitate over. It will change as a result.
These are testable hypotheses, not doctrine.
What we believe.
Material evidence matters.
Objects are not educationally powerful by magic. Their distinctive potential lies in giving learners evidence that can be examined, questioned and reasoned from.
Objects are starting points for enquiry.
They should generate questions and investigation, not merely illustrate messages decided in advance.
Observation comes before explanation.
Learners need time to notice, describe, compare and wonder before interpretation is supplied for them.
Claims must answer to evidence.
Interpretation and imagination matter, but learners should distinguish what evidence supports, what it complicates and what remains uncertain.
Story should serve enquiry.
Narrative can create connection and curiosity, but should hand the learner back to evidence rather than close interpretation down.
Mediation makes encounter educational.
Objects become powerful through skilled questioning, dialogue, handling, close looking and scaffolding.
Authenticity and provenance must be made legible.
Whether something is original, replica, reconstruction or generated should be visible, and the significance of that distinction should itself be open to enquiry.
Uncertainty is productive.
Fragments, absences, ambiguity and competing interpretations are not failures to eliminate. They are opportunities to practise judgement and revision.
Inclusion must be designed into enquiry.
Multisensory, embodied and dialogic approaches should create multiple routes into the same evidential problem.
Technology should extend human judgement, not displace it.
Digital and AI tools earn their place when they deepen access, comparison, questioning or reflection while leaving the learner's intellectual work with the learner.
What signing asks us to do.
We will make evidence visible.
We will identify originals, replicas, reconstructions and generated material, and be open about provenance, uncertainty and interpretation.
We will create enquiry, not simply deliver content.
Learners will have genuine opportunities to notice, question, infer, test and revise.
We will mediate skilfully.
We will treat questioning, dialogue, handling and close observation as professional practices requiring thought, skill and development.
We will design for participation.
We will build sensory, linguistic, physical and cultural access into encounters from the beginning rather than retrofit inclusion afterwards.
We will use story and technology transparently and purposefully.
We will use them to open, deepen and extend enquiry, not to obscure the status of evidence or perform the learner's thinking for them.
We will transmit the pedagogy.
We will treat object-based enquiry as professional knowledge to be taught at entry, honed in practice, evaluated, shared and refreshed across a career.
The lines we will not cross.
We refuse to treat authenticity as educational magic.
Real does not automatically mean learning.
We refuse to reduce objects to props for predetermined stories.
If the learner has nothing left to investigate, the object has been closed down.
We refuse to blur evidence, interpretation and invention.
Learners should be able to tell what is known, inferred, reconstructed and generated.
We refuse to confuse fluency, novelty or engagement with understanding.
An impressive output or a captivated audience is not evidence that learning has happened.
We refuse to optimise uncertainty and intellectual friction out of learning.
Difficulty, ambiguity, disagreement and revision are part of becoming capable of judgement.
The point of education is not to eliminate friction.
Tell us where this breaks. That is more useful than agreement.
What the sector actually needs next.
This is the one part of the manifesto with no answer in it. These are the possibilities, not a predetermined package. Practitioners at the GEM Conference will be asked which of them would be most useful to their own practice, and which their organisation could realistically invest time or money in. Those two answers are often not the same, and the difference is the useful part. There will be more than one way to have a say: the session itself, this site, a wider consultation in the autumn and an online session.
Evidence and sector paper
The renewed case, evidence-informed and publishable.
Manifesto and advocacy tool
Refined through consultation and openly endorsed, for use in making the case.
Practical framework or toolkit
Enquiry routines, authenticity and replicas, inclusion, AI and digital use, curriculum links.
Training and CPD
A route by which this pedagogy is taught, honed and refreshed across a career.
Evaluated pilot
Tested in real museum and school settings, with evidence rather than assertion.
Research and evidence programme
Filling the gaps this proposition has exposed rather than assumed.
Network or community of practice
Somewhere the practice is shared, challenged and kept alive.
Something else
Something the sector needs that is not on this list.
There is a further question underneath. An entry-level module for someone new to the work, a practice development route, a refresher for experienced staff, or something built for freelancers and casual staff. These are not the same thing and they are not equally provided for.
Nobody has the whole picture. That is the point.
This is a working model, not a finished framework. It is organised by what the learner is doing rather than by what kind of material is in front of them.
Encounter
What is actually here?
Slow down. Observe. Handle or inspect where appropriate. Describe before explaining. And establish what kind of source this is: original, replica, photograph, reconstruction, testimony or generated representation.
Question
What might this tell us?
Notice patterns. Wonder. Compare. Infer. Connect to what is already known. Generate questions. Story can create the way in here.
Test
What does the evidence support?
What is directly observable? What is inference? What do we know from provenance or context? What has been reconstructed or imagined? What other evidence would we need? This is the stage made newly urgent by synthetic media.
Revise
What do I think now?
Make a claim. Hear another interpretation. Test it. Change your mind where you should. Name what remains uncertain. That is judgement.
It loops. A reconstruction sends you back to the fragment. An AI output gives you a claim to test. A new piece of context makes you look again.
None of these is automatically educational. Across all four stages runs mediation: questions, dialogue, oracy, handling, scaffolding, time, attention and feedback. And across the whole framework runs inclusion: sensory access, language and literacy, physical access, cultural responsiveness, choice and agency, and multiple ways to participate.
The object plus mediation is the intervention.
Use whatever strengthens the learner's encounter with evidence and capacity for judgement. Be explicit about what each source is, what it can tell us, and what it cannot.
Not simply because of AI.
Reform is strengthening enrichment and disciplinary thinking.
Source: Curriculum and Assessment Review, 2025 · Enrichment Framework, 2026
A framework is coming, and object enquiry is built out of talk.
Source: Every Child Achieving and Thriving, 2026
Policy insists that technology should support human judgement rather than replace it.
Source: DfE, Generative AI in Education, 2025 · Lee et al., CHI 2025
Now carries much greater explicit weight in inspection.
Source: Ofsted inspection framework and toolkits, 2025 to 2026
All of this sits inside childhood increasingly shaped by screens and synthetic media.
Source: DfE and DHSC screen use guidance, 2026, and in development
These currents do not prove the argument. They make the educational questions behind it much harder to ignore.
The case for learning from objects has not faded. It has sharpened.
Sources: DfE, Curriculum and Assessment Review: final report, November 2025. DfE, The enrichment framework for schools and colleges, June 2026. DfE, Every child achieving and thriving, February 2026. DfE, Generative AI in education, 2025 onwards. Ofsted, inspection framework and toolkits, 2025 to 2026. DfE and DHSC, screen use guidance for children under five, March 2026, with guidance for children aged 5 to 16 announced June 2026 and expected autumn 2026. Lee et al., The Impact of Generative AI on Critical Thinking, CHI 2025, Microsoft Research and Carnegie Mellon University.
The first open session with the sector, on day two of the GEM Conference. Practitioners will rank the principles, discard one, audit the commitments against their own organisations and mark the refusals they hesitate over.
A logic model and theory of change, built from what that room produced. Published here, and sent to everyone who endorses the approach.
Wider consultation and an online session. Whatever the sector ranked highest is what gets tested next.
Back to GEM with evidence rather than opinion.
Whatever the sector decides it actually needs. That question is deliberately still open.
Endorse the approach today. Help rewrite it. Then help decide what the sector actually needs from it.
This is not a petition, and you are not signing a finished document. Endorsing means you back the direction and want a part in working out what it becomes.
You back the direction and want to be kept informed.
You want to challenge the principles, commitments and refusals, and see the drafts before they are settled.
You want a say in what gets built, and possibly a place in the advisory group.
One short form. Around two minutes.
The Real Thing is developed by Adam Goldwater, ASG Goldwater Consulting, with the support of GEM. It goes to the sector for the first time at the GEM Conference 2026 at Bletchley Park. The proposition is in development and open to challenge.
The ‘real thing’ is not simply the object. It is the real intellectual work of learning: encountering evidence, asking questions, testing claims and exercising judgement. It is the encounter between a learner, evidence, and a question that has not already been answered for them.