AI adoption: are rare cases slipping off the agenda?

Starting AI adoption with recurring tasks seems reasonable. The more precise management question is how to prevent that initial priority from permanently sidelining rare but important situations.

What the literature establishes

James March (1991, Organization Science) formalises the tension between exploiting existing knowledge and exploring new possibilities. His models show how more immediate and predictable returns can favour exploitation at the expense of exploration. Applied to AI, this framework invites managers to examine not only the gains from selected uses, but also the problems to which the team stops paying attention.

The mechanism: learning mainly from what recurs

Daniel Levinthal and James March (1993, Strategic Management Journal) describe a myopia of learning: organisations may privilege what is nearby and successful while neglecting distant horizons and failures. Feedback on a repetitive activity is more plentiful than feedback on an exceptional situation. A portfolio of AI uses could therefore concentrate on what quickly produces evidence of progress, without covering what matters most when an incident occurs.

The premature conclusion

It would be excessive to conclude that rare cases should take priority or be entrusted to AI. Beginning with a frequent, bounded activity can support learning and make results observable. The qualification concerns selection: low frequency alone is not sufficient grounds for removing a problem from the management agenda. (our executive and employee training programmes)

What these studies do not demonstrate

These articles offer models and a theoretical analysis of organisational learning, not trials of generative AI in Geneva-based teams. They establish neither the optimal distribution of uses nor the situations that a current system can handle reliably. A high usage count indicates repetition of a practice, not coverage of important situations.

A practical check in Genève

As part of SHR’s programme « Conduire l’adoption de l’IA dans son équipe », a Geneva-based team can draw up a list of infrequent but high-stakes situations before selecting its uses: an atypical private banking case, an exceptional request in an international organisation, an unusual defect in luxury watchmaking or a documentation anomaly in commodity trading. After one month, calculate the proportion of these situations with a documented decision: AI use to be tested, human handling retained, or review deferred with a stated reason. The numerator is the number of situations with a decision and the denominator is the fixed initial list; the register remains internal and requires no sensitive data to be sent to an AI tool. This measure checks the coverage of management decisions, not the tool’s performance. To go further: explore the Leading AI Adoption in Your Team training in Geneva in the canton of Geneva, or browse our executive and employee training programmes in Switzerland.

In pictures: Leading AI Adoption in Your Team in Geneva

• leading ai adoption in your team
• Geneva
• research
• training geneva