INMA Researcher-in-Residence Greg Piechota posed a thought-provoking question to attendees at INMA’s Media Innovation Week in Antwerp this week: What are we going to do when we stop playing with ChatGPT?

Despite the tremendous noise about AI’s disruption, the real impact is still to be determined by how customers adopt the technology, he cautioned: “It’s always customers that disrupt businesses, not tech alone.”

The role of news media companies is to ask how the industry can create value, strategise, manage risks, and organise this disruption.

Use AI as a synthesis tool

Piechota has spent the last six months researching the uses and potential impacts of generative AI across the media industry, and had shared some of the insights he has learned.

AI is creating a content tsunami that will flood the news market, and the number of AI-generated news sites has already ballooned. Piechota presented data that showed one AI-generated news site output 1,200 articles per day compared to 100 from Le Monde or 250 from the New York Times.

There is an opportunity for news media in the potential premium on journalism amid this flood of misinformation.

Piechota reminded the audience that people have talked about information overload since Gutenberg. But technology that makes it easier to create, publish, and distribute content also leads to developing  tools to help people synthesise and manage the information overload.

“Maybe generative AI is not what we think. It’s not about generating things, it’s actually about synthesising things. If this is the case we’re going to find tremendous disruption.” 

News still needs a human touch

Despite efficiency gains with AI, Piechota’s research shows there is still a dependency on human intervention for credible journalism. “Journalism is a process — verification, investigation, witnesses, sense making. You need to understand that you will always have a cost for these processes behind the content.”

Fact-checking is one task that cannot be automated, and it comes at a high cost. The fact-checking cost per story has been calculated to be as much as US$470 for a publisher in Germany and almost US$3,000 for a publisher in the U.K.

Piechota believes the financial gains from AI may be overestimated. Rather than asking how to do the same work with fewer resources, the introduction of AI is an opportunity to ask how to do more with the same resources. “The goal is not to have fewer people. It’s about having people on what really matters and automating the rest.”

With the introduction of AI, new guidelines are needed to help manage the risks. Employees need clear policies and training on AI tools to help them adapt to working alongside machines.

Piechota cautioned the audience against assuming news media can make the same gains with AI as big tech. There are risks to news media: “Big tech doesn’t hang on trust and presence of truth.”

Focusing on what matters

The potential for AI is enormous but news media has limited time and limited resources. Piechota said news media companies need to prioritise when incorporating AI into the newsroom. “The focus should not be what can generative AI can do, but what needs to be done.”

If implementing AI is like building a house, nobody is going to wait for the entire structure to be built. “Nobody can wait for the results that long,” Piechota said. News media needs to prioritise solutions that offer return on investment. Taking the analogy further, Piechota suggested that companies focus on their direct needs: If you need to sleep, build a bedroom and if you need to eat, build a kitchen.”

Piechota cited several case studies of AI application in his presentation, including the case of Le Monde, which expanded into new markets with its English edition based on automated translations of 40 articles per day. Five people work on selection and verification of translations: “They have a real product.” 

Johannes Gorset, Director of Engineering for VG (Norway) joined Piechota onstage for a panel discussion following the presentation. Both speakers talked about the possibilities for AI in their presentations, which led to the inevitable question: Where do we start?

Piechota advised not to ask what AI can do for you, but rather where AI could make the biggest difference in things that matter. “We’re likely to see too many initiatives. We need to focus on prioritisation. Don’t jump into the first opportunities. Think about it.”

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