AI: define the problem before asking for solutions
AI tools can quickly generate options for addressing delays, non-conformities or coordination difficulties. The management question is specific: how can teams prevent the initial wording submitted to the tool from becoming the official definition of the problem without scrutiny?
What the literature establishes
Amos Tversky and Daniel Kahneman (1974, Science) describe anchoring: in estimation tasks, a starting point can influence judgement even when it has little relevance. Adjustments made from that starting point may remain insufficient. This finding invites scrutiny of where reasoning begins; it does not demonstrate that an initial AI response necessarily imposes its framing on a team.
The mechanism that sustains an initial framing
Raymond S. Nickerson (1998, Review of General Psychology) reviews research on confirmation bias: the search for and interpretation of evidence can favour an existing belief or hypothesis. Asking AI how to address a “lack of motivation” thus presupposes a diagnosis that the facts may not support. The management risk considered here is that an abundance of solutions makes that assumption less visible.
The premature conclusion
One might conclude that problems must always be defined without AI. These studies do not justify such a prohibition: the tool can also help generate competing formulations and make their assumptions explicit. The useful distinction is between producing solutions within a given frame and examining whether that frame should be adopted. (our executive and employee training programmes)
What the evidence does not establish
These publications address human judgement, not current uses of generative AI in companies in Fribourg. They measure neither the frequency of incorrect framing with these tools nor the effectiveness of a protocol designed to prevent it. Obtaining several reformulations is not sufficient either: they may all retain the same untested causal hypothesis.
A practical check in Fribourg
Within SHR's “Managing in the Age of Artificial Intelligence” programme, an exercise could use an anonymised case of food-industry non-conformity or delays in a bilingual industrial company, potentially with methodological review from Fribourg's university community. Before requesting solutions, the team would record the problem in French and German, the observed facts, the assumed causes and an observation that could contradict each cause. Over a pilot period defined in advance, the measure would be the proportion of cases containing these elements before the first request for AI-generated solutions, verifiable through dated records and conversation histories. This indicator would document disciplined problem formulation, without being presented as proof of better decisions. To go further: explore the Managing in the Age of Artificial Intelligence training in Fribourg in the canton of Fribourg, or browse our executive and employee training programmes in Switzerland.
In pictures: Managing in the Age of Artificial Intelligence in Fribourg
- managing in the age of artificial intelligence
- Fribourg
- research
- training fribourg