AI adoption: should we continue because we have already invested?
After buying a tool and training the team, discontinuing an AI use case can feel like repudiating a collective decision. The management question is specific: how can we decide whether to continue without turning past investment into an obligation to persist?
What the literature establishes
Barry Staw (1976, Organizational Behavior and Human Performance) experimentally studied escalating commitment in a resource allocation decision. In that setting, personal responsibility for the initial decision encouraged further investment following negative outcomes. The finding highlights a management difficulty: the person who selected a solution is not necessarily best placed to reassess its funding dispassionately.
When past effort becomes an argument
Hal Arkes and Catherine Blumer (1985, Organizational Behavior and Human Decision Processes) describe the sunk cost effect: having already invested money, time or effort can encourage continuation of a course of action. Their research highlights, in particular, the desire not to appear wasteful. Applied to AI, this mechanism suggests a hypothesis to test rather than an established finding: training efforts might become an argument for maintaining a use case whose future value remains uncertain.
The premature conclusion
One might conclude that every use case should be discontinued as soon as initial results disappoint. That would confuse escalating commitment with persistence justified by benefits that remain plausible. An acquired skill can reduce future costs; an expenditure that has become irrecoverable, however, is not in itself a reason to continue. (our executive and employee training programmes)
What these findings cannot settle
These studies concern neither generative AI nor workplace teams in Valais. They provide no universal threshold for deciding when to stop, and continuing after a failure does not in itself demonstrate bias. The assessment must also consider exit costs, contractual commitments and the value of further learning that can still be expected.
A practical check in Sion
As part of the programme “Leading AI adoption in your team”, a team in Sion could apply this distinction at its next review of AI use cases. Whether the application involves documentation in hydropower, administrative tasks in healthcare, commercial content in viticulture or visitor information in Alpine tourism, each use case would be assessed by separating sunk expenditures, future costs and expected benefits. The proposed measure is the proportion of continuation decisions whose written record specifies a verifiable future benefit, a deadline for checking it and an alternative considered, including discontinuation. This indicator checks the traceability of the decision, not AI’s profitability; it nevertheless makes decisions justified solely by “we have already invested” identifiable. To go further: explore the Leading AI Adoption in Your Team training in Sion in the canton of Valais, or browse our executive and employee training programmes in Switzerland.
In pictures: Leading AI Adoption in Your Team in Sion
- leading ai adoption in your team
- Sion
- research
- training sion