Why this works
Students Do Not Need a Finished Idea to Begin
Many creative classes ask a student to know what they want before they are allowed to start. That is the moment where confidence breaks. Hypersketch turns it around: they make something rough, and use what comes back to find the idea.
They do not even need a hunch to earn permission to explore. They explore to find the hunch.
A way in, always
A doodle, a few shapes, a fragment. Enough to make part of their thinking visible — that is the only entry requirement.
Out of the obvious box
They are answered with several readings of their mark, including ones they did not intend — so the first predictable idea is not the only one on the table.
Judgement stays with them
Nothing that comes back is an instruction. The student decides what is interesting, what is weak, and what it means.
Heard in classrooms
“Is this correct?”
“What are we supposed to make?”
“I don’t know what to draw.”
“Can I just use the first one?”
None of these mean a student lacks creativity. They mean the student is waiting for permission.
How many decisions the student makes
Answer-first AI
Prompt → polished answer → selection
1 decision
Visual Co-Thinking
Just Make → Provoke → Notice & Shortlist → Connect Dots ↻ then Reframe → Build & Branch → Communicate
6+ decisions, all discussable
AI produces possibilities. It does not do the thinking. That distinction is the whole method — and it is the one thing worth protecting in your classroom.
The category
Visual Co-Thinking
AI that helps students form intent, not merely execute it.
Most visualisation tools
“Tell me what you want, and I will help you produce it.”
Hypersketch begins earlier
“Make something incomplete, and let us discover what it could become.”
Visual Co-Thinking means making part of a thought visible, seeing what it suggests, and using what emerges to create new associations, questions and possibilities. It lets thoughts form through interaction rather than requiring them to be fully articulated in advance.
This is why Hypersketch does not behave like a perfectly obedient image generator. A useful provocation keeps enough of the student’s input to preserve authorship, while introducing enough variation to create discovery — and it should not close the thought too quickly.
You do not need an idea to start creating. You create in order to discover the idea.
The shape of a session
The Possibility Loop™, End to End
A repeatable way to turn uncertainty into expanding possibilities.
Phase 1 repeats for as long as it needs to. Only when a student spots something worth pursuing does the track leave the loop and run down into Phase 2.
Each possibility becomes a new place to begin.
This is not a shortcut around thinking. It is a scaffold for beginning, connecting, reframing and continuing to create.
Stage by stage
What to Say, and What to Watch For
A few prompts per stage — pick whichever fits the room. A question usually moves students on better than an explanation.
Phase 1 — Provocation
Students begin without a clear idea — a doodle, a few shapes, a rough form.
“Make something you cannot explain yet.”
“Make 10 weird and wonderful toys.” (when the brief is not about toys)
“Use the ‘Surprise Me’ button.”
“Draw and connect 2 random shapes.”
Watch for: Students who sit waiting for an idea before they touch anything.
Hypersketch answers their marks with a field of possible readings — not one authoritative answer.
“Draw an unexpected line.”
“Leave some ambiguity.”
“Use a wrong prompt or shape.”
“What did it show you that you did not intend?”
Watch for: Students treating an output as the answer instead of as exploratory material.
The student looks for what is surprising, strange, useful or unresolved — and shortlists many.
“What do you see here that you were not expecting before?”
“What looks like it could do something new?”
“What just attracts you?”
“What might be a happy accident?”
Watch for: Students who are avoidant of shortlisting any imperfect or incomplete visual.
New associations form between what appeared and what the student already knows or cares about.
“What could this go with? Who else might need this?”
“Where does this look like it would be used?”
“Who could use this, and to solve what problem?”
“What is this unlikely to be?”
“What if this became a different object?”
Watch for: Students who are trying to conjure only flawless ideas.
Phase 2 — Possibility
The connection changes the question. This is the one stage with no AI — the judgement is entirely the student’s.
“Has the problem you started with changed?”
“Say the opportunity in one sentence.”
“Who is this for now, and why them?”
Watch for: Students defending their first topic rather than embracing shifts.
They develop the promising direction while still opening alternatives.
“Vary it — and show us five other ways it could go.”
“What would the bolder/simpler version look like?”
“Which part is worth keeping if the rest changes?”
“Which part is unnecessary and could be removed?”
Watch for: Committing too early. Branching is not indecision.
Fidelity goes up and the proposition goes in front of other people.
“Say only one problem this solves.”
“Say why it is different or better.”
“Show it in a more explanatory view.”
“Show it in use, show it in context.”
“Show it moving.”
Watch for: Presentation and image quality replacing reasoning. Ask for the rationale, not for pretty pictures.
How to judge the work
Do Not Assess Only the Final Images
A gallery of polished outputs is weak evidence of learning. The strong evidence is the visible development of the student’s thinking — and the loop leaves a trail of it.
Ask of any project
What did the student begin with?
How many genuinely different possibilities were considered?
What unexpected detail did the student notice?
What new connections did the student form?
What did the student reject, and why?
Which ideas were combined?
How did the original problem change?
What further branches were explored?
What decision did the student make independently?
What can the student explain now that was not apparent at the start?
Where did the student exercise judgement rather than accept an AI response?
What a full trail looks like
Initial marks → first provocations → what they noticed → connections and annotations → rejected and combined directions → the reframed idea → the developed concept → further branches → the eventual direction.
On drawing
Drawing stays in the process. What changes is how much proficiency a student needs before drawing becomes useful to them.
From “can you draw this beautifully?” to “can you make enough of your thought visible for new possibilities to emerge?”
Rough drawings, incomplete forms, doodled lines, simple shapes, imperfect marks, partial objects — all of it is workable. The goal is not to help students avoid drawing. It is to help more students use drawing as a way to think.
The final image matters. The evolution of the thinking matters more.
Who built the method
Developed Through the Practice of Teaching and Creating
The Hypersketch curriculum, application and creativity mechanism were developed by Associate Professor Donn Koh, who teaches design innovation and artificial intelligence at the National University of Singapore.
A recipient of the NUS Outstanding Educator Award, he has mentored student product-creation projects that have collectively achieved more than 300,000 units in global sales and received multiple James Dyson and Red Dot Design Awards.
References to Donn Koh’s academic appointment are included solely for identification and context and do not imply endorsement by the National University of Singapore.
“How can people begin thinking creatively before they have the confidence, drawing proficiency or fully formed intention normally expected of a designer?”
The answer was not an easier image generator. It was an environment where more people could enter the creative process and exercise real judgement inside it.
Participation
Built for Students Who Do Not Yet See Themselves as Creative
Many creative interfaces quietly tell beginners they do not belong. A blank canvas suggests strong drawing skills are required. A prompt box suggests a clear idea and the right words are required. A polished output suggests unfinished work is inferior.
Hypersketch lowers the cost of beginning. Students enter through small marks, rough drawings and partial thoughts — they are not asked to perform expertise before they are allowed to explore.
Not just a new way for confident creative students to work.
A new way for more students to become creative participants.
Especially for students who say
“I cannot draw well enough.”
“I am not creative.”
“I don’t know what to make.”
“My idea is probably wrong.”
“Other students are naturally more imaginative.”
These are the students the Singapore Polytechnic cohorts were mostly made of.
The evidence · Singapore Polytechnic
Three Cohorts, Mostly Non-Design Students, One Large Shift
Singapore Polytechnic's Academic Quality & Research unit measured students before and after they learned this process. We did not run the study ourselves.
d = 0.7
How much more creative students believed they were, after learning the process.
On the yardstick researchers use for this (Cohen's d), 0.7 is a big move — the kind you would notice in the room, not only in a spreadsheet.
They did not just make more. They started to see themselves differently.
Educator access