AI adoption: does fluent writing seem more reliable?
Generative AI tools make structured, professional-looking texts readily available, without that presentation guaranteeing accuracy. The management question is specific: how can a team avoid treating ease of reading as evidence of reliability?
What the literature establishes
Rolf Reber and Norbert Schwarz (1999, Consciousness and Cognition) demonstrated experimentally that perceptual ease of reading could influence judgments about whether statements were true. Their manipulation concerned visual readability, not the quality of an argument or an AI-generated output. Nevertheless, the finding establishes an important point: judgments about content can depend on a feature of its presentation.
Processing ease as information
Adam L. Alter and Daniel M. Oppenheimer (2009, Personality and Social Psychology Review) review research on processing fluency: the subjective experience of ease with which information is processed. This experience can contribute to various judgments, but its meaning depends on the context and how it is interpreted. For a team manager, the mechanism suggests a hypothesis worth examining: a well-written response might be accepted for reasons that extend beyond its supporting evidence.
The hasty conclusion
It would be excessive to conclude that clear writing is suspicious or that AI responses should be made harder to read. Clarity remains a communication asset; it simply does not constitute factual validation. The task is to separate two assessments: “this document is understandable” and “its claims are sufficiently supported”. (our executive and employee training programmes)
What the evidence does not establish
These articles measure neither generative AI adoption nor decisions made by teams in Lausanne. Moving from a readability manipulation to the assessment of a professional document is an extrapolation whose extent needs to be tested in practice. A satisfaction questionnaire about response quality cannot therefore establish, on its own, whether a team distinguishes convincing presentation from substantiated content.
A practical check in Lausanne
As part of SHR’s programme “Conduire l'adoption de l'IA dans son équipe”, a Lausanne team in medtech, higher education, international sport or a Lake Geneva region scale-up could work with a public or fictional case file containing no sensitive data. Prepare two versions with exactly the same claims and references but differing in presentation fluency, then randomly assign them to readers who see only one version and identify the claims they consider sufficiently supported. The measure is the percentage-point difference between the proportions of claims accepted in each condition, distinguishing documented claims from undocumented ones. This exploratory exercise would not demonstrate a general AI effect, but it would make any team sensitivity to presentation observable. To go further: explore the Leading AI Adoption in Your Team training in Lausanne in the canton of Vaud, or browse our executive and employee training programmes in Switzerland.
In pictures: Leading AI Adoption in Your Team in Lausanne
- leading ai adoption in your team
- Lausanne
- research
- training lausanne