As news organisations look for ways to engage younger audiences, a new study from the Center for Media Engagement examines whether using AI to tailor stories for Gen Z can help. The results highlight the challenges of using generative AI to reach young adults and reveal broader issues with AI use in news that extend beyond Gen Z.

Study set-up

Researchers used AI to rewrite published articles into two styles: one aimed broadly at Gen Z and another targeting people aged 18 to 21 years old. Both versions featured short, digestible articles emphasising the importance of the issue for young people.

The Gen Z version also used informal language and Internet slang. All articles where AI was used included an AI disclosure below the byline.

To understand how people react to these AI-revised articles, an online experiment was conducted with half of the Gen Z participants and half of the older participants. They were assigned to read either the original article (where no AI was used), the article revised by AI for Gen Z, or the article revised by AI for 18- to 21-year-olds.

After reading the articles, participants were asked what percentage of the article they thought was written by an AI editor. They were also asked what they thought about the article, including if they thought it was accurate, trustworthy, and enjoyable to read.

Separate from the article they read, participants were asked to rate how acceptable they found several newsroom uses of AI.

Results

Readers did not evaluate the AI-edited articles more positively than the traditional articles. Both Gen Z and older readers had similar impressions of the articles, intentions to engage, and learning outcomes, no matter which version they read.

In addition to the muted response toward the AI-edited articles, people’s reactions to perceived use of AI provide a warning for newsrooms. The more AI people thought they detected, the less favourably they felt about the article, the lower their intentions to engage with it, and the lower their score on a quiz of what they learned from the article.

Moreover, when asked about their attitudes toward AI practices in journalism, people of all ages were generally opposed to the use of AI without human oversight. There was, however, more support for AI use with oversight, especially among older participants.

Assumptions about journalist AI use

People thought AI was used to write articles, even when it wasn’t. Overall, only 14% of participants thought no AI was used, even when reading an original article.

The reader’s age did make a difference in how much of the article they believed was written by AI: Gen Z readers thought just over one-third of the article, on average, was written by AI. Older participants thought just over half of the article, on average, was written by AI, even when reading the original version.

Another difference between age groups is how they detect AI. No matter which type of article they read, including the one with no rewrites by AI, older participants’ perceptions of AI use remained the same.

Gen Z readers, however, detected significantly more AI when they read the article rewritten for Gen Z compared to the article rewritten for 18- to 21-year-olds and the original article. The Gen Z version of the article used informal language and Internet slang, suggesting Gen Z readers form impressions about AI based on the article’s tone.

An important takeaway for all age groups is that AI disclosure labels were easily missed. Reading an article with an AI disclosure label made no difference in how much of the article people thought was written by AI.

Takeaways for newsrooms

The results raise concerns about the use of AI personalisation in news stories.

Using this approach to reach Gen Z may not be effective and could even turn readers away. As an alternative, newsrooms might consider reaching younger audiences by including their perspectives and stories in an article.

Since readers of all ages often assume articles are at least partially written with AI (and the more AI people perceive, the less favourable the response), journalists should consider explaining where AI is or is not being used in their work.

It might also be valuable to clearly lay out the news organisation’s policies and boundaries on AI use. If newsrooms are looking for guidance, Poynter and Trusting News offer useful starting points.

Additionally, given how often readers overlook AI disclosure labels, journalists may need to explore more noticeable ways to signal AI use.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .