Lectures and conversations

When the use of AI fails: Evidence from a survey experiment on public perception of trust and acceptability in the news media

Emilie Midtsæter Strønen, PhD Candidate at the Department of Information Science and Media Studies, presents results from a survey experiment about the use of AI in articles from news media and public administration.

Emilie Midtsæter Strønen is a PhD candidate at the Department of Information Science and Media Studies. In this presentation, she will show results from a factoral experiment from the Norwegian Citizen Panel, measuringnews organizations' and public administration's use and failed use of AI can affect perceptions of trust and acceptability.

Light lunch will be served, as first come, first served. 

The event is hybrid, if you can not join us in the Corner room at Sofie Lindstrøms hus, you can join us digitally. (external link)

 

Welcome!

Abstract

AI is currently being used in the news media for various tasks, including in the production of news articles. This study examines how news organizations’ use of AI and failed use of AI can affect public perception. A 2 (authorship: human vs. AI) x 2 (outcome: no error vs. error) x 2 (actor: news media vs. public administration) factorial experiment was conducted (N = 2030). Results showed evidence of an AI-penalty on perceived trust and perceived acceptability, as well as a strong error-penalty. AI-errors appears to be penalized more strongly than human-errors. Comparing the results for the news media sector to AI-written reports in public administration, I observe similar patterns. However, several of the penalties appear to be significantly stronger for public administration.  

Zoom link for digital participation

Contact person

Anne Marthe Borgen