Audiences demand human oversight of AI in news, studies find
Two studies in Digital Journalism examine audience expectations for AI use in news. A conjoint experiment in Chile (Valenzuela et al.) found human review of all AI content is the most influential factor for credibility and news selection. Disclosure of generative AI use also matters, but audiences showed no preference regarding AI for menial tasks or personalization. A qualitative study (Zier & Diakopoulos) revealed that audiences want AI labels to ensure accountability and trust, distinguishing between 'generated by' and 'assisted by' AI. Participants worried about hallucination and bias, demanding human oversight. However, labels can backfire, prompting fact-checking or skepticism. The studies suggest publishers should emphasize human involvement and carefully craft labels.
Key facts
- Human oversight of all AI content was the most influential factor for credibility in a Chilean conjoint experiment.
- Disclosure of generative AI use was important for credibility and news use.
- Audiences showed no preference for AI use in menial tasks or personalization.
- Participants distinguished between 'generated by' and 'assisted by' AI labels.
- AI labels can backfire by prompting fact-checking or skepticism.
- The studies were published in Digital Journalism.
- Authors include Sebastián Valenzuela, Ingrid Bachmann, Porismita Borah, Natalia Solís Valdés, Jessica Zier, and Nicholas Diakopoulos.
- The research was covered by Nieman Journalism Lab on June 17, 2026.
Entities
Institutions
- Nieman Journalism Lab
- Nieman Foundation for Journalism at Harvard University
- Digital Journalism
- University of Washington
- University of Texas at Austin
- UCLA
Locations
- Chile
- United States
- United Kingdom
- Germany
- Seattle