AI: who keeps operational rules up to date?
When a team asks an AI assistant to draft responses or instructions, it often supplies procedures assumed to represent the work. Within SHR — Swiss Human Resources’ programme « Manager à l'ère de l'intelligence artificielle », this raises a specific question: who decides that a rule needs to change, and how does that change reach the tool?
What the literature establishes
Martha Feldman and Brian Pentland (2003, Administrative Science Quarterly) distinguish the general representation of a routine from its actual performances by particular people in particular situations. Their theoretical analysis shows that the relationship between these aspects can generate stability, but also flexibility and change. A written procedure should therefore not be mistaken for the entire routine it seeks to describe.
How work changes the rule
Martha Feldman (2000, Organization Science) draws on a field study to describe how the outcomes of action can lead organisational members to modify their routines. Gaps between what was expected and what actually happens become opportunities for revision, rather than merely anomalies. For an AI assistant drawing on procedures, this suggests a challenge distinct from the initial quality of the documents: organising their revision in light of experience.
The premature conclusion
One might conclude that the tool simply needs to be continuously fed with observed practices. That would confuse appropriate adaptation, temporary workarounds and unauthorised departures from the rules. The manager’s role is to ensure these differences are examined and validation is assigned to the appropriate people, not to turn every deviation automatically into a new rule. (our executive and employee training programmes)
What these studies do not demonstrate
These articles concern organisational routines, not contemporary AI assistants; they measure neither their reliability nor the effect of updating frequency. Their contribution is an analytical framework, not proof that a particular document management process improves performance. In practice, replacing a document does not itself establish that the tool now uses the correct version: that requires a separate test.
A practical check in Sion
In Sion, a team can select a procedure used by its assistant: preparing a hydropower maintenance intervention, arranging a healthcare appointment, receiving grapes during the harvest or providing access information for an Alpine tourism activity. For one month, it records every identified gap between the reference procedure and operational needs, with an explicit decision: revise, retain or authorise an exception. The measure is the proportion of approved revisions for which an accountable owner, an effective date and a documented test of the assistant’s response are all present; if no revision is approved, the indicator is reported as not calculable. This exercise can provide a practical application of the SHR programme, using fictional or sanitised cases without directly changing any safety-critical operational instruction. To go further: explore the Managing in the Age of Artificial Intelligence training in Sion in the canton of Valais, or browse our executive and employee training programmes in Switzerland.
In pictures: Managing in the Age of Artificial Intelligence in Sion
- managing in the age of artificial intelligence
- Sion
- research
- training sion