AI: what can we learn from pioneer teams alone?
An AI rollout is often planned around a handful of internal experiences considered promising. The management question is specific: can an organisation decide to scale up when only teams still running their experiments present results?
What the literature establishes
Jerker Denrell (2003, Organization Science) uses theoretical analysis and simulations to show that the underrepresentation of failures can distort learning from others’ experience. Observing surviving organisations can make practices appear beneficial even when they are not beneficial across the full population. Applied to AI case reports, this mechanism suggests examining how the cases were selected, not just how convincingly they are described.
How early results shape what follows
James G. March (1991, Organization Science) examines the tension between exploring new possibilities and exploiting existing knowledge. His model shows how the more immediate and predictable returns from exploitation can lead organisations to favour it excessively, at the expense of longer-term learning. In an AI rollout, one hypothesis to test is therefore that early, visible applications attract resources while less conclusive trials disappear from the comparison.
The premature conclusion
It would be excessive to conclude that pioneer teams necessarily give a misleading picture of AI. Their experience can document valuable conditions for success without establishing that those conditions exist elsewhere. The important distinction is between evidence that a use case is feasible and evidence that it can be transferred to other teams. (our executive and employee training programmes)
What the available evidence cannot tell us
Neither article studies generative AI or organisations in Lausanne: they offer mechanisms of organisational learning, not an estimate of the success of a local rollout. In practice, a collection of case reports becomes difficult to interpret when the total number of trials, discontinued experiments and resources committed are undocumented. Even a complete inventory cannot establish that the tool caused the results if teams differ in their available time, expertise or support.
A practical check in Lausanne
As part of SHR’s « Manager à l'ère de l'intelligence artificielle » programme, an exercise applicable in Lausanne is to inventory every trial of a particular use case over a period defined in advance, including both ongoing and discontinued trials. Whether in a medtech company, a higher education institution, an international sports organisation or a scale-up in the Lake Geneva region, the register should record the task, resources committed, outcome and reason for continuing or stopping. One verifiable measure is the coverage rate of the dossier submitted to the decision-making committee: the number of trials documented in it divided by the total number identified, cross-checked against access requests and project registers. This rate does not prove AI’s effectiveness; it checks whether the decision rests solely on experiences that still have a public-facing success story. To go further: explore the Managing in the Age of Artificial Intelligence training in Lausanne in the canton of Vaud, or browse our executive and employee training programmes in Switzerland.
In pictures: Managing in the Age of Artificial Intelligence in Lausanne
- managing in the age of artificial intelligence
- Lausanne
- research
- training lausanne