AI: does a repeated claim become evidence?
A strategy note, a presentation and a summary produced with AI can repeat the same claim without adding any evidence. The precise management question is: how can this circulation across documents be prevented from being mistaken for agreement among independent sources?
What the literature establishes
Lynn Hasher, David Goldstein and Thomas Toppino (1977, Journal of Verbal Learning and Verbal Behavior) showed that repeating statements could increase their perceived validity. A previously encountered claim can therefore seem truer without becoming better supported. The finding concerns judgements of truth, not the objective quality of information.
Knowledge does not necessarily neutralise repetition
Lisa K. Fazio, Nadia M. Brashier, B. Keith Payne and Elizabeth J. Marsh (2015, Journal of Experimental Psychology: General) observed an illusory truth effect even for statements that participants had the knowledge needed to recognise as false. Familiarity can therefore influence judgement without available knowledge being consistently brought to bear. For a manager, the audience's expertise is not enough to guarantee that every repetition will be reconsidered.
The premature conclusion
It would be excessive to conclude that repeating information is always harmful, or that several AI-assisted summaries necessarily make a team credulous. Repetition remains useful for communicating instructions; it simply does not, on its own, provide new evidence. The relevant distinction is between multiple documents and multiple genuinely independent sources of factual support. (our executive and employee training programmes)
What these studies cannot establish
These experiments concern neither generative AI nor executive committees in Zurich. They do not measure the effect of document chains in which a summary feeds into a presentation and then a recommendation. Applying their findings to these practices provides a reason for vigilance, not a demonstrated estimate of local risk.
A practical check in Zurich
Within SHR's « Manager à l'ère de l'intelligence artificielle » programme, one exercise involves examining the key claims in a decision dossier at a Zurich financial services, insurance or technology company, or at the European headquarters of an international group. For each claim, the team records the documents repeating it, its original factual source and any independent corroborating evidence; an untraceable origin remains explicitly flagged. The measure is the proportion of claims appearing in multiple documents for which no additional independent support is identified, reported alongside the total number of claims examined. This record does not measure whether those claims are false: it checks whether the dossier actually distinguishes repetition from corroboration. To go further: explore the Managing in the Age of Artificial Intelligence training in Zurich in the canton of Zurich, or browse our executive and employee training programmes in Switzerland.
In pictures: Managing in the Age of Artificial Intelligence in Zurich
- managing in the age of artificial intelligence
- Zurich
- research
- training zurich