Last year, MIT’s Media Lab came out with a buzzy study on a phenomenon that’s been coined “cognitive debt.” In short, the researchers found that when ChatGPT users offloaded essay writing tasks to their AI assistants, their own writing abilities declined. ChatGPT users consistently had the lowest “brain engagement” and task performance scores across the study.
The research sparked debates about whether AI assistants can contribute to the atrophy of other skills, too, not just essay writing. That’s where a new study in Journalism Practice picks up. Shangyuan Wu, a researcher at the National University of Singapore, dug into the question of whether journalists are “deskilling” as a result of AI adoption. She also asks what skills journalists may be learning or developing further — in other words, “upskilling.”
The report’s authors interviewed 14 working journalists in Singapore, a hub for AI development in Asia and home to a mix of local, regional, and international newsrooms. The participants were pulled from print, broadcast, online outlets, and wire services.
One of the main concerns in these interviews was that journalists may lose foundational research skills. Given that generative AI products can quickly produce basic background reports and conduct online searches, some interviewees speculated that these tools could contribute to journalists losing the skills needed to “track down information” and check primary sources. The study also suggested that easy access to background reports on a topic could mean reporters have less motivation to conduct expert interviews or otherwise speak to people to better understand a topic.
Overall, the study surfaced concerns from working journalists that AI tools were “dumbing down the profession,” as one interviewee put it, because AI-generated research, interview questions, and story drafts could erode journalists’ ability to think for themselves.
Some interviewees in the study pushed back on these ideas, drawing parallels to past technological innovations like AI transcription tools and search engines. One interviewee said that AI just helps to “streamline research processes,” framing AI technologies as efficiency tools rather than replacements for manual research and reporting.
The study also outlines several journalistic muscles that will be flexed more due to AI adoption, suggesting there is potential for “upskilling” in these areas. For one, the authors point to the “ability to discern.” Since AI tools can regularly output hallucinated, biased, or incomplete information, journalists will need to strengthen their ability to discern “between fact and fiction, facts and opinions, good and bad journalism or writing, and biased and unbiased content.”
Journalists will also need to bolster their fact-checking and editing abilities, some interviewees suggested. If journalists are handling more AI-generated copy, they will need to make sure that copy is factually accurate and doesn’t read as generic or stilted. Some also posited that AI tools are not good at “storytelling” and that the information provided by them will still need to be tied together by a journalist in a compelling or convincing way.
Zooming out, the study points to the ability to come up with “original ideas” as an increasingly vital skill for journalists as AI-generated text, audio, and video floods our information ecosystems.
“AI can give you what is out there, but it cannot provide that innovation or creativity,” said one interviewee. “Humans are still the ones required to ‘push the boundaries.’”
The full study outlines several other potential areas for deskilling and upskilling in journalism. You can read it in Journalism Practice.
