AI-assisted committees: does shared information crowd everything else out?
When a committee prepares a decision using an AI-generated summary, its members have a common starting point. The management question is specific: how can this shared foundation be prevented from overshadowing information that only particular functions can contribute?
What the literature establishes
Garold Stasser and William Titus (1985, Journal of Personality and Social Psychology) studied group decisions in which participants did not all hold the same information. Their experiments show that discussion favours information known to everyone over information held by just one member, even when the latter could support a better decision. Bringing together people with complementary information is therefore not enough to pool their knowledge.
The mechanism: more opportunities to return to familiar information
Garold Stasser and William Titus (1987, Journal of Personality and Social Psychology) examined how information load and the proportion of shared information influence the dissemination of unshared information. Their work illuminates an asymmetry: information known to several participants has several opportunities to be mentioned, whereas exclusive information depends on its sole holder. A shared AI-generated summary might reinforce this common starting point; this is a proposed application of the findings, not an effect measured in these studies.
The premature conclusion
One might conclude that shared summaries should be abandoned or that everyone should prepare entirely separately. These studies do not support that conclusion: a common foundation can also clarify criteria and prevent misunderstandings. The issue is to distinguish what is already shared from what still needs to enter the discussion, rather than mistake initial agreement for effective knowledge pooling. (our executive and employee training programmes)
What these data cannot establish
These experiments concern laboratory tasks with a controlled distribution of information, not industrial committees using generative AI. In practice, information held by only one person is not necessarily accurate, relevant or decisive. Measuring its presence in the discussion tells us about knowledge circulation, not directly about decision quality.
A practical check in Neuchâtel
In a Neuchâtel company working in microtechnology, watchmaking or precision microelectronics, an industrialisation review could test a structured round of contributions from each function before any decision based on an AI summary. After reading the summary and before the meeting, each representative separately records verifiable facts from their field that are missing from it; the minutes then allow the team to calculate the proportion of these facts explicitly examined before the decision. Comparing this proportion across comparable reviews, with and without the structured round, would provide an indicator of information circulation, without on its own proving that decisions have improved. Within SHR’s « Manager à l’ère de l’intelligence artificielle » programme, this protocol would constitute a verifiable management exercise rather than a promise about the tool’s performance. To go further: explore the Managing in the Age of Artificial Intelligence training in Neuchâtel in the canton of Neuchâtel, or browse our executive and employee training programmes in Switzerland.
In pictures: Managing in the Age of Artificial Intelligence in Neuchâtel
- managing in the age of artificial intelligence
- Neuchâtel
- research
- training neuchâtel