IA on the job: can a rejected option change the choice?
An assistant can prepare a comparison table of suppliers, equipment or tourism services. The management question is precise: would the recommendation remain the same if a clearly less attractive option were removed?
What the literature establishes
Joel Huber, John W. Payne and Christopher Puto (1982, Journal of Consumer Research) showed that adding an option dominated by one offer, but not by the other, could increase the probability of choosing the offer that dominates it. This option provides a favourable point of comparison without needing to be selected itself. A comparison list is therefore not necessarily a neutral container: its composition can help shape preferences.
The mechanism: a preference that is easier to justify
Itamar Simonson (1989, Journal of Consumer Research) examined attraction and compromise effects from the perspective of choice justification. An option can become more attractive because it seems easier to defend against the alternatives, particularly when it occupies an intermediate position. Applied to AI, this mechanism suggests examining not only the written arguments, but also the set of options within which those arguments become persuasive.
The premature conclusion
One might conclude that the list should always be reduced to the highest-ranked options. These studies do not justify that rule: removing a genuinely relevant possibility can also impoverish the decision. The useful distinction is between an option that introduces a real trade-off and one that mainly makes its neighbour look more appealing. (our executive and employee training programmes)
What these studies cannot establish
These studies concern experimental consumer choices, not professional decisions supported by generative AI. They establish neither the frequency nor the magnitude of these effects in a company in Sion. A preference shift after changing the list is therefore a signal to investigate, not automatic proof of manipulation or a poor decision.
A practical check in Sion
Within SHR’s « IA on the job » programme, a possible exercise in Sion would compare maintenance offers for hydropower, non-clinical equipment in healthcare, bottling services in viticulture or booking services for alpine tourism. Using verified documentation, show two randomly assigned groups either two offers or those same offers accompanied by a third, less favourable than one of them on the selected criteria and with no identified compensating advantage, keeping all other information and instructions unchanged. Measure the difference between the proportions of participants choosing each original offer, then ask for a justification based on criteria defined before the exercise. This local test does not demonstrate a general effect, but it can establish whether the composition of the comparison deserves scrutiny before an actual decision. To go further: explore the AI on the Job training in Sion in the canton of Valais, or browse our executive and employee training programmes in Switzerland.
In pictures: AI on the Job in Sion
- ai on the job
- Sion
- research
- training sion