Imagination, experimentation are key to generative AI in newsrooms

By Peter Bale


New Zealand and the U.K.


Artificial Intelligence — and its uses and risks in journalism — was the stand-out topic at the Perugia International Journalism Festival, and it was far from a gloomy outlook with leading practitioners seeing opportunities for newsrooms to embrace the technology and innovate to create better journalism that engaged with audiences.

Gina Chua, the executive editor at Semafor, reminded the Perugia audience what large language models were good for and what they were not intended to do.

Speakers at the International Journalism Festival spent a lot of time discussing generative AI and its risks in journalism.
Speakers at the International Journalism Festival spent a lot of time discussing generative AI and its risks in journalism.

“They are not fact models. They are not verification models. What they do is they do language incredibly well,” she told a panel.

She gave the example of experimenting with stories and asking ChatGPT to change the style and tone to that of a different publication, the racy New York Post.

“The key point is I wasn’t asking for facts. [I asked it to] take this single story and rewrite it in different ways. It did not introduce error because it was working on a story,” she said. “We can think about how to regenerate news and reach people in ways we haven’t even thought of.”

A good example of this is the series of experiments journalists at Schibsted ran.

Semafor was using AI chatbots to proofread stories, Chua said. The tools had the potential to generate new story ideas and story types, to fire imagination in the newsroom and deliver new products to readers. For average journalists, it could offer a competitive advantage to improve the quality of their writing and the impact of their work.

“We desperately need more imagination … we need to stay current because we need to understand how it’s being used elsewhere,” she said. “One of the most interesting things we don’t talk about enough is this a tool that will both level up and create a different kind of competition … . This could improve their output dramatically.”

In the same session, Lisa Gibbs, director of news partnerships at the Associated Press, dealt with the question of the potential threat to journalism jobs in the emerging era of generative AI.

“I have always said that AI will not replace journalists, but they make really good assistants. And I think today we are poised even more to realise this vision of AI truly being an assistant,” she said, referring to existing AP use of Artificial Intelligence to analyse company earning reports and experiments the organisation is running to extract information from documents.

Lisa Gibbs, director of news partnerships at the Associated Press, spoke at the festival.
Lisa Gibbs, director of news partnerships at the Associated Press, spoke at the festival.

Gibbs suggested that journalists experiment with the various models but take extreme care not to go live with what they truly didn’t fully understand.

“I am concerned that the rush to sound smart and try things with ChatGPT is leading journalists to rush into some things they don’t have the knowledge of,” she said. Apart from the question of inaccurate or “hallucination” in some of the tools, the question of copyright and where responsibility might lie for incorporating copyrighted material into articles derived from AI.

She suggested the industry needs a framework to understand how to apply generative AI.

“We need a framework to help identify when should we be using automation and AI. We really need to make sure we are applying the same values and ethical frameworks to the use of these technologies as we have been to transcription and automatic stories for years,” she said.

See also AP CEO warns of “existential threat” to publishers from generative AIThe Economist covered the same session, calling out the “remixing” of journalism that generative AI would allow — creating a new “soup.”

Chris Moran, head of editorial innovation at The Guardian, said newsrooms needed to get familiar fast with the tools their journalists were probably already using: “Even if you’re an editor who thinks you hate this stuff ... your editors and staff may be doing this already.”

In a separate sessionLondon School of Economics professor Charlie Beckett — whose Polis think tank has been working on Artificial Intelligence for years — said generative AI had the potential to free journalists from labourious tasks and challenged them to do better work. It offered relief from “boring repetitive labour you don’t want to do as a journalist, weather and sports results [for example]. Your time is better spent doing something else.”

In one of the closing sessions at a conference rather obsessed with Twitter and the evident loss of credibility on the platform under the ownership of Elon Musk, Emily Bell, director of the Tow Center for Digital Journalism at Columbia University, said Twitter under Musk was no longer a safe space for journalists. As a public company, Twitter had worked to understand journalists.

“They understood what it was to be a journalist, what it means to have pro-democratic platforms … . It is no longer a safe platform for journalism,” she said. “It’s vandalism of infrastructure and that is what dictators do as well.”

The next International Journalism Festival in Perugia is scheduled for April 17-24 2024. 

If you’d like to subscribe to my bi-weekly newsletter, INMA members can do so here.

About Peter Bale

By continuing to browse or by clicking “ACCEPT,” you agree to the storing of cookies on your device to enhance your site experience. To learn more about how we use cookies, please see our privacy policy.