An Italian newspaper says it has created the world’s first issue generated entirely using AI, with journalists limited to asking questions of a chatbot and reading the answers before inserting them. The Guardian earlier reported on the initiative by Il Foglio, an Italian liberal conservative paper.
Claudio Cerasa, editor of Il Foglio, said the experiment serves as a test for how AI could work “in practice” in a newsroom and forces journalists to ask tough questions about the technology’s impact on the industry.
“It will be the first daily newspaper in the world on newsstands created entirely using artificial intelligence,” said Cerasa. “For everything. For the writing, the headlines, the quotes, the summaries. And, sometimes, even for the irony.”
The four-page “Il Fogolio AI” was inserted into the larger Tuesday edition, and can also be viewed online.
Early experiments with using generative artificial intelligence in newsrooms have not gone particularly well. Back in 2023, CNET faced criticism after it quietly began publishing financial advice stories generated using AI which turned out to include significant inaccuracies. More recently, the Los Angeles Times released an AI-powered tool called “Insights” that was supposed to rate the bias of opinion articles automatically generated counterpoints; it quickly pulled the tool after it was found downplaying the KKK.
Generative artificial intelligence is good at producing verisimilitudes of genuine writing, something that looks clear and authoritative. There have been attempts to improve the “thinking” process of chatbots, but they are ultimately glorified autocomplete systems and face the intractable problem of simply making things up. Chatbots that present their logic as they produce a response will even sometimes admit as much. Ultimately, the problem with all language models is that the user has to look closely at all the generated text and correct errors if they even spot them at all. Newsrooms in particular have to be careful to not harm their credibility amongst the public even further by publishing slop.
Still, news organizations continue to experiment with the technology despite significant concerns amongst journalists in particular regarding whether newsrooms will try to use AI to cut back on staffing. Patch, a hyperlocal news site that was previously owned by AOL, now relies entirely on AI scraping the web to find news for many of its local editions.
Quote-unquote “real” journalism should be less affected, as generative AI is simply creating new text from material it has already seen before. Original reporting—finding original stories, interviewing individuals—requires producing entirely new information that is not yet on the web. But the public at large does not place much value on media in the digital age—not just limited to journalism, they are not willing to pay much for music or video anymore either. Labor being the highest cost center in most organizations, it is not hard to see more news organizations using AI wherever possible.
Someday, we are going to have AI news articles citing information from other AI news articles until one cannot even figure out the original source anymore. Or stories generated entirely from Reddit comments.
