IA on the job: reusing past success without locking in old practices?
In SHR — Swiss Human Resources' “IA on the job” programme, using internal knowledge raises a question that goes beyond document retrieval. Before adopting a solution suggested by AI from a successful case, how can a manager check whether that precedent remains relevant?
What the literature establishes
Barbara Levitt and James G. March (1988, Annual Review of Sociology) describe organisational learning as the encoding of lessons from experience into routines that subsequently guide action. Their review highlights that organisations can learn from experiences whose causes and outcomes they interpret imperfectly. A case retained as a success therefore provides a record of action, not proof that its method suits every comparable situation.
The mechanism: what works becomes easier to repeat
James G. March (1991, Organization Science) analyses the tension between exploiting existing knowledge and exploring new possibilities. His model shows how the more immediate and predictable benefits of exploitation can favour it at the expense of exploration. Applied to an assistant that retrieves and reformulates precedents, this mechanism suggests a hypothesis: making old solutions more accessible could encourage their reuse even when circumstances have changed.
The hasty conclusion
It would be excessive to conclude that an assistant drawing on internal archives prevents innovation. Reusing an established method can avoid unnecessarily repeating work and free up resources to examine what has actually changed. The useful distinction is less between old and new than between a precedent whose conditions of application have been checked and one adopted simply because it is available. (our executive and employee training programmes)
What these studies cannot establish
These articles offer a theoretical review and a model of organisational learning; they do not measure the effects of today's generative assistants. They therefore cannot quantify a risk of organisational rigidity or establish that assisted document retrieval reduces exploration. In practice, an archive that retains the solution but omits constraints, abandoned attempts and the conditions behind success makes precisely this check difficult.
A practical check in Fribourg
As a practical application of “IA on the job” in Fribourg, a team in the agri-food sector or bilingual industry, potentially supported by a partner from the local university community, can test the reuse of precedents on non-critical cases. Before each reuse, it records the source case reference, one condition necessary for applying it to the new situation, and current evidence with which to check that condition, whether in French or German. The indicator is the proportion of precedent-based proposals for which this condition was checked before approval, retaining both the numerator and denominator. This record measures disciplined transfer, not an increase in innovation; unchecked cases indicate where to seek additional information or try another approach. To go further: explore the AI on the Job training in Fribourg in the canton of Fribourg, or browse our executive and employee training programmes in Switzerland.
In pictures: AI on the Job in Fribourg
- ai on the job
- Fribourg
- research
- training fribourg