Creativity with AI: better proposals, but fewer options?
Generative AI can quickly produce starting points for a product, a service or an industrial problem. The managerial question is more precise: how can it be used without the team’s proposals becoming increasingly alike?
What the literature establishes
Anil R. Doshi and Oliver P. Hauser (2024, Science Advances) studied a short-story writing task, with or without access to ideas generated by AI. Access to these ideas improves creativity ratings, but the AI-assisted stories are more similar to one another. The finding therefore distinguishes the quality attributed to each output from the diversity of the collection.
Another mechanism: organising idea generation
Michael Diehl and Wolfgang Stroebe (1987, Journal of Personality and Social Psychology) studied productivity losses in group brainstorming. Their experiments identify production blocking — having to wait while someone else speaks — as an important explanation. Their work does not address AI, but it shows that the contribution process, not just individual ability, determines which ideas become available.
The premature conclusion
One might conclude that AI should be excluded from creative sessions. The story-writing study does not justify that rule: it identifies an individual benefit alongside lower collective diversity in a particular task. Gathering independent ideas before consulting a shared AI proposal is a working hypothesis to test, not a procedure validated by these studies. (our executive and employee training programmes)
What the evidence cannot promise
Short stories are neither food formulations nor industrial maintenance solutions. Their similarity does not directly predict the economic value of options, their feasibility or their diversity within a bilingual team. Assessing only the proposal ultimately selected also leaves any narrowing of the alternatives explored out of view.
A practical check in Fribourg
Within SHR’s « Manager à l'ère de l'intelligence artificielle » programme, an exercise in Fribourg could bring together managers from the food sector and bilingual industry, with methodological support from the local university community. For comparable, non-confidential problems, alternate individual idea collection before consulting AI with collection after a shared AI proposal has been presented, keeping time and the number of contributions equal. Have bilingual assessors who do not know the production condition code the French and German proposals using a predefined classification of solution families. The measure would be the number of distinct families for an equal number of proposals: a verifiable indicator of local diversity, not general proof that one method is superior. To go further: explore the Managing in the Age of Artificial Intelligence training in Fribourg, or browse our executive and employee training programmes in Switzerland.
In pictures: Managing in the Age of Artificial Intelligence in Fribourg
- managing in the age of artificial intelligence
- Fribourg
- research
- training fribourg