09 University of Michigan / arXiv · revised 2025 paper. ([arxiv.org](https://arxiv.org/abs/2401.13481))

AI can widen the idea pool—and still create imitation pressure

A large dynamic experiment examined how exposure to AI-generated ideas affects creativity, diversity and adoption over time.

3,000+

responses from participants in 48 countries

The Story

This large experiment looked at creative ideation as a chain: participants saw ideas from earlier participants, then generated their own. Some chains included AI-generated examples and some did not.

The researchers found that high AI exposure increased global idea diversity in their experiment, while average creativity scores did not differ significantly between conditions. The result is more nuanced than the usual ‘AI makes ideas better’ claim.

For practice, the design implication is to vary the inputs. Use AI for provocation, then deliberately introduce human references, lived experience, constraints and critique so the studio’s output does not converge around the same convenient defaults.

Studio Wensday Angle

Use AI to diversify prompts, not to outsource taste.