AI on the job: more ideas, or more variations?
Within SHR — Swiss Human Resources’ “IA on the job” programme, using AI to prepare a brainstorming session raises a specific management question. Should generated proposals be displayed at the start of a workshop, or should participants’ independent ideas be collected first?
What the literature establishes
Michael Diehl and Wolfgang Stroebe (1987, Journal of Personality and Social Psychology) investigated why brainstorming groups produce fewer ideas than people working separately whose proposals are subsequently pooled. Their experiments identify production blocking — notably the inability to speak simultaneously — as a major explanation for this loss. This finding concerns how participation is organised, not the effectiveness of an AI assistant.
When an example becomes a frame
David G. Jansson and Steven M. Smith (1991, Design Studies) showed, in design tasks, that exposure to an example could lead participants to reproduce some of its features, including problematic ones. A visible proposal therefore does more than supply content: it can also shape what is imagined next. An AI-generated list displayed at the outset could play this role, but that extension remains a hypothesis.
The premature conclusion
It would be excessive to conclude that AI should be excluded from creative workshops. Instead, the two studies suggest distinguishing two problems: enabling everyone to contribute and preventing a single example from framing every contribution. A workshop can solve the first through individual writing while retaining the second if everyone starts from the same generated list. (our executive and employee training programmes)
What the evidence cannot promise
These studies examine neither today’s generative assistants nor organisations in Berne. They do not establish that an independent contribution will be better, more feasible or more original in every professional setting. Counting proposals is not sufficient either: several formulations may describe exactly the same change.
A practical check in Berne
As part of “IA on the job”, a proposed exercise in Berne could compare two formats using a fictional case: an AI list displayed before participants contribute, or individual contributions collected before the same list is displayed. The case could address reception services in the federal administration, a public health campaign, telecoms support or a handover in precision manufacturing, without using sensitive data. With comparable session lengths and group sizes, have anonymised proposals classified by assessors unaware of the format, using a professional assessment framework agreed beforehand, then count the distinct solution categories and those judged feasible. The observed difference is a local indicator to replicate, not proof that one sequence is universally superior. To go further: explore the AI on the Job training in Bern, or browse our executive and employee training programmes in Switzerland.
In pictures: AI on the Job in Bern
- ai on the job
- Bern
- research
- training bern