IA on the job: how many options should a team be given?
Within the “IA on the job” programme offered by SHR — Swiss Human Resources, one management question deserves to be separated from idea generation: how many options should be presented to the team responsible for deciding? In Berne, this applies equally to choosing an information initiative in federal administration or public health and to selecting an operational response in telecommunications or precision manufacturing.
What the literature establishes
Sheena S. Iyengar and Mark R. Lepper (2000, Journal of Personality and Social Psychology) showed across several experimental settings that an extensive choice can attract more interest without encouraging action. Depending on the experiment, smaller assortments notably encouraged purchasing or engagement with a task. These findings distinguish two phenomena that AI use can blur: appreciating a rich selection and managing to choose.
An effect that does not appear everywhere
Benjamin Scheibehenne, Rainer Greifeneder and Peter M. Todd (2010, Journal of Consumer Research) examined choice overload in a meta-analysis. They found an average effect close to zero, alongside substantial variation between studies. Their work therefore calls for identifying the conditions under which an extensive list hinders decision-making, rather than treating that difficulty as an automatic consequence of the number of options.
The premature conclusion
One might conclude that AI should always be asked for a short list. That would turn a conditional effect into a general rule: a smaller selection can also eliminate a relevant solution before the team has examined its criteria. The management question therefore concerns the transition from exploration to selection, not a universal cap on options. (our executive and employee training programmes)
What the evidence does not allow us to transfer
These studies address neither lists generated by generative AI nor professional decisions in organisations in Berne. A consumer decision does not require the same justification as a public health or industrial process decision. In practice, the number of options often varies along with their relevance, redundancy and presentation quality: attributing every difficulty to volume would obscure these differences.
A practical check in Berne
For a practical application of the “IA on the job” programme in Berne, SHR could propose a trial using simulated cases without sensitive data, drawn from federal administration, public health, telecommunications and precision manufacturing. Starting from the same pool of verified options, randomly assign participants either a full list or a shortlist compiled using explicit criteria, with an identical descriptive format. Measure the proportion of decisions made within the allotted time that meet a professional assessment rubric established before the trial and applied by someone unaware of which version participants received. The result will help evaluate this selection process, without claiming to isolate the effect of option count alone or establish an ideal length for every decision. To go further: explore the AI on the Job training in Bern in the canton of Bern, or browse our executive and employee training programmes in Switzerland.
In pictures: AI on the Job in Bern
- ai on the job
- Bern
- research
- training bern