When AI becomes unavailable: who keeps the work going?
AI integration is usually discussed in terms of its uses and results, less often in terms of its absence. The useful question is specific: which activities can the team still carry out if the tool becomes inaccessible, using what resources and under whose responsibility?
What the literature establishes
Susan Leigh Star and Karen Ruhleder (1996, Information Systems Research), drawing on a study of scientific computing infrastructure, show that infrastructure develops through its relationship with existing practices, conventions and resources. Its operation tends to become transparent in use, while breakdowns reveal the dependencies it supports. Applied to AI, this perspective invites managers to examine not just the tool, but the activities that have gradually ceased to function without it.
An outage can also disrupt roles
Karl E. Weick (1993, Administrative Science Quarterly) analyses the Mann Gulch disaster as a collapse of shared meaning and role structure within a firefighting team. His analysis highlights the importance of resources that allow coordination to be rebuilt when familiar reference points disappear. Without equating an IT outage with that disaster, it raises an organisational question: who allocates work when the tool that structured the flow of activities is no longer accessible?
The premature conclusion
One might conclude that a manual procedure must be retained for every use of AI. These studies do not justify such a rule: some activities can be postponed, while others require an alternative. The task is to decide explicitly what must continue, what can wait and who can authorise operation at a reduced service level. (our executive and employee training programmes)
What the evidence cannot promise
These studies do not concern generative AI and do not measure the effectiveness of a continuity plan when it becomes unavailable. They illuminate organisational mechanisms without providing a universal threshold for preparedness. Within SHR’s “Managing in the Age of Artificial Intelligence” programme, they provide a starting point for distinguishing a written procedure from a capability that has actually been practised.
A practical check in Sion
In Sion, an exercise using fictional cases could involve a hydropower maintenance request, a healthcare administrative schedule, a wine order or an Alpine tourism booking enquiry. Without interrupting actual services, the team handles the case without AI, with a designated fallback coordinator and time and quality criteria established beforehand. The measure is the proportion of all tested cases completed within the agreed time and meeting those criteria; each blockage is recorded alongside the missing resource. This result does not certify organisational resilience, but it checks whether the promised continuity exists for the activities examined. 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