AI Co-Creation Surveys: How Participants Can Build Better Ideas

An AI co-creation survey asks people to make something, not merely evaluate something made for them. A participant might start with a real-life context, describe what would improve it, generate a first visual expression, then react to and refine that expression. The task can explore a product, service, experience, message, identity or future scenario.

This approach is useful when a team needs to understand the shape of an idea as well as its acceptance. A rating can signal preference. It rarely reveals the details someone would add, remove, protect or personalise. Co-creation makes those choices visible and gives the participant a more active role in forming the material under discussion.

Whalets are Whale’s mobile-first AI co-creation surveys. They make the process accessible through a sequence of simple, conversational tasks, so people do not need design training or prompt-writing expertise to contribute.

A finished concept gives participants a narrow set of ways to respond: they can like it, dislike it, or explain a reaction. Creating starts from a different premise. It gives someone room to surface the context, language, imagery and trade-offs that sit behind their response.

For example, a participant visualising a better public space may choose to add trees, shade or room to pause. Those choices are not a final design brief, but they give researchers something concrete to explore: what need does the change address, what feeling should the space create, and which details seem most important?

This is consistent with a broader finding from MIT Sloan reporting that generative image tools can help creative workers move beyond familiar patterns by rapidly making possibilities visible. In research, the key is to use that visualisation as a conversation starter and evidence trail, not as an automated answer.

The best tasks are structured enough to reduce the blank-page problem but open enough to preserve the participant’s point of view. Begin with one focused research question and lead people through a small number of choices. Each step should clearly explain what is being asked and why it matters.

Prompt scaffolding is especially important. Rather than asking for one perfect prompt, help participants build towards it: their situation, the role the idea should play, the feeling it should create, and the details they want to see. They still make the creative decisions; the structure simply gives them a route into the task.

The final visual is only one part of the response. Review the original intention, first prompt, emotional reaction, instructions for change, what was rejected, and the participant’s explanation. Together, these interactions can show both the desired outcome and the priorities that shaped it.

Do not over-interpret a single generated image. Look across participants for recurring needs, language, emotional cues and trade-offs. Keep the participant’s context with the material so that visual similarity does not get mistaken for shared meaning.

A well-designed co-creation survey therefore adds depth to research; it does not replace sampling, facilitation, analysis or human judgement. The goal is richer participant expression and a clearer route from that expression to the team’s next question.