Follow us to keep up to date with future research, and let us know which insights you found the most interesting! 💬

Some findings:
→ Confirmed: AI disclosures increase reader skepticism
→ Users enjoy "surprise" news recommendations & diversity in their feeds
→ News outlets differ in language & prioritisations of AI risks depending on national & political contexts
→ Scenario use aids in tech research outreach

Surprise me! A longitudinal user study on serendipitous interface design in news recommender systems
by Zilin Lin, @damiantrilling.net, Stuart Duncan, Kasper Welbers, and @susanvermeer.bsky.social
🔗 ceur-ws.org/Vol-4027/pap...

Informing AI Risk Assessment with News Media: Analyzing National and Political Variation in the Coverage of AI Risks
by Mowafak Allaham, @kimonkieslich.bsky.social, and Nicholas Diakopoulos
🔗 doi.org/10.1609/aies...

Scenarios in Computing Research: A Systematic Review of the Use of Scenario Methods for Exploring the Future of Computing Technologies in Society
by @juliabbarnett.bsky.social, @kimonkieslich.bsky.social, Jasmine Sinchai, and Nicholas Diakopoulos
🔗 doi.org/10.1609/aies...

“Transparency is More Than a Just Label": Audiences’ Information Needs for AI use Disclosures in News
by @hannescools.bsky.social, S. Morosoli, @laurensnaudts.bsky.social, K. Venkatraj, @claesdevreese.bsky.social, & @natalihelberger.bsky.social
🔗 osf.io/preprints/so...

📅Deadline: November 1st, 2025
🔗Read More: werkenbij.uva.nl/en/vacancies...

A big thank you to Laura and Shannon for starting this conversation in our Lab! We're looking forward to seeing this research getting published and informing professional decisions in journalism.

💬 Because this is still a new phenomenon, the strength of these reactions to AI may change over time and across publics, especially if new developments emerge. Audiences may become accustomed to AI use in media production, or further negotiate how they approach the issue from a moral standpoint.

🔎 AI disclosures caused Gen Z to go into "detective mode": once they noticed the AI label, they stopped focusing on the news content and instead tried to find visual flaws. Any initial trust in the post was lost upon seeing the label. So, transparency can backfire and distract in big ways.

🏷️Without labels, AI-generated images got rated as more credible than those taken by photojournalists — participants had trouble identifying whether the images were artificial or not. This suggests that image generator performance is good enough to fly under the radar, even in journalistic contexts.

Combining info from eye trackers with participant interviews, they analyzed how young people process news in Instagram post formats, what they notice first, and how they interpret different types of image credit labels (taken by a photojournalist, created with AI, or no attribution at all).