Information is felt to be more credible when it is read out than when it is formatted as text. A study into the credibility of ‘smart’ speakers such as Alexa in comparison to Google search results and Wikipedia articles1 shows that even inaccurate or false information is regarded as far more credible if it is read out by a language assistant. An experiment presented information on appendicitis, first of all giving the commonest symptoms. The next sentence stated that about 80 percent of those affected do not have the typical symptoms. Contradictions and misinformation like this were recognized less often if they were conveyed in speech.
Two processes play a part in this phenomenon: Firstly, through social signals such as a human-like voice, the dialog with a language assistant gives us the feeling of interacting with an intelligent being. So, as in a conversational situation, we assume that what is said is true, without asking after sources or additional information. Secondly, it’s more difficult to process spoken language. Written text can be reread; it’s easier to find inconsistencies. As Internet users, we’ve learned to check sources of information. Likewise, we should be conscious that dialog-oriented interaction with language assistants make them appear more credible, and we should always apply critical thinking to the new technologies.
Prof. Dr. Sonja Utz, Leibniz-Institut für Wissensmedien (IWM)
1Gaiser, F., & Utz, S. (2024). Is hearing really believing? The importance of modality for perceived message credibility during information search with smart speakers. Journal of Media Psychology: Theories, Methods, and Applications, 36(2), 93–106. https://doi.org/10.1027/1864-1105/a000384
