Phase 1 · Explore

From Object to Character

Pick a museum object that pulls you in. Investigate it slowly, with eyes only — no AI yet. Ask whose voice the museum left out. Then, and only then, let the AI guess what that missing person looked like — and put the guess away for later.

Duration · ~60 minutesSteps · 4Output · Character description & Day One image

1. Pick an object from the museum

Pick ONE object that feels interesting, surprising, beautiful, uncomfortable, or important.

The object screen — two questions before you commit

Before you fall in love with an object, check both. Both answers must be YES:

☐ Is there an identifiable human role attached? A maker, user, trader, servant, owner — someone whose hands touched it.
☐ Can something be set against the story we are already given? Is there an account (label, textbook, catalogue) this object lets you question?

If either answer is no, the object will not sustain your investigation — however beautiful it is. Pick another one.

Capture the basics

Object name — what is it called on the museum label?
Object description — a few sentences in your own words.
Quick sketch — even a rough drawing helps you look more carefully.

2. Investigate your object

Get to know your object. Be curious. Investigate it further. Do not use AI yet. Do not start the story yet. Just look.

Write at least 5 things you can see

Stick to what is literally visible. Examples of openers:

It is made of…
It is the colour…
It is big / small…
It has symbols…
It looks old / broken / shiny / heavy…
It looks beautiful / ugly / strange because…

What does the museum NOT tell us?

Read the museum label or exhibition text. Pick the most useful information. Now ask:

Who used this object?
Who made it?
How did people feel about it?
Who benefited from it?
Who is missing from this story?

3. Ask: Whose story is missing?

Now think about people connected to the object. Who could be missing? Who is not visible in the museum story?

For example: maker, worker, child, woman, trader, servant, traveller, outsider, religious person.

The four corners

For your object, brainstorm a person in each of these four roles:

Who used it?	Who benefited?
The everyday users — the people whose hands actually touched this object.	The people who gained — wealth, power, status, comfort — because this object existed.

Who made it?	Who suffered?
The makers — workers, craftspeople, artisans who produced it. Often anonymous.	The people who paid the cost — those whose labour, freedom, or wellbeing was taken to make this object exist.

Dot vote

Each team selects one character linked to the object to tell an "untold story." 2 dots per team.

CHECKIs your missing character realistic?

1. ASK — Ask someone who may know: a teacher, museum guide, museum staff, or expert.
2. SOURCE IT — Find 2 sources. For each one, name three things: who made it · when · where it was published. If you cannot name those three, you have not found a source yet — Google, a chatbot and a search box are doors, not sources. AI output is never a source.
3. CHECK SIDEWAYS — Leave the source: open a new tab and see what other people say about whoever made it. Then compare your two sources — do they agree? If not, write what is different.
Final check · Can this person really be linked to the object, time, and place? If unsure, write: "We are not sure because…"

4. The Day One image — the machine's guess

Now — and only now — ask the AI to picture your person. Do not describe them. Just say who they are. The AI will fix their age, gender, clothing, health and wealth in one go — from its own assumptions, before you know anything. That is exactly the point. This image is the machine's guess, not your character. You will not use it in your story — you will hold it up against the evidence at the end of Phase 4.

AGenerate the guess

One generation. No retries, no improvements — a guess does not get to be polished.

Day One prompt · paste into your AI image tool

Picture [persona], a [role] in [time/place]. That is all the information you get. Do not ask me questions — just show me the person.

BLabel it and put it away

Save the image. Write on it: "The machine's guess — before we knew anything."
Do not use it in your story. Not as a reference, not as a starting point.
You will bring it back for the Two Portraits comparison at the end of Phase 4 — next to an image built from evidence.

CRead the guess — 5 minutes

Before you put it away, look at what the machine decided for you. Note at least three assumptions it made:

Look at: age · clothing · health · wealth · skin · expression · setting · who else is in the picture at all.

Where do those assumptions come from? Keep your notes — they are evidence for the AI Audit later.

DComplete your character

Template · fill in for your character

I am a __________________________________________ I live in / come from ____________________________ My connection to the object is: __________________ This perspective is interesting because __________

LOW-AI ROUTE

Skip the AI image — draw your own guess instead, before the evidence work starts. Label it and keep it the same way. The comparison at the end of Phase 4 works just as well against a drawn guess.

CHECKIs your Day One image doing its job?

1. NO DESCRIPTION — Did you tell the AI only who the person is — not their age, clothing, health, or appearance?
2. LABELLED — Is "The machine's guess — before we knew anything" written on the saved image?
3. THREE ASSUMPTIONS — Did you note at least three things the machine decided on its own?
Final check · Is the image put away — and NOT used in your story? You will meet it again at the end of Phase 4.