“We need to prepare for the death of the article as we know it,” Nikita Roy said at the Nordic AI in Media Summit (NAMS/#NAMS25) in Copenhagen recently.

For me, this was probably the most strikingly simple, yet clarifying statement made over the course of the two days — speaking, as it does, to a number of implications of the impact of AI on journalism.

It’s about tech, but it’s also very much about people — those who produce it and those who consume it. (Roy, founder of Newsroom Robots Lab, spoke at #NAMS25, a not-for-profit event organised by the Nordic AI Journalism network and hosted by JP/Politikens Hus. The presentations are freely available on the NAMS YouTube channel.)

To understand the implications for newsrooms and users, it’s important to understand how far AI (generative AI, in the main) has come in just the last couple of years.

At the first edition of this AI Summit back in 2023, the focus was largely on AI tools helping journalists be faster and better at the same tasks they were already performing. Incremental development of existing workflows, in other words.

Today, AI can do almost everything technical involved in the production of a piece of journalism without much involvement of a journalist.

Except one thing. It cannot find the story and it cannot connect with the people involved in it.

What does this have to do with the death of the article? To my mind, the biggest takeaway of the AI conference is that journalism does not reside in formats like articles, videos, and podcasts. Journalism is the story, the data, the context and conclusions.

While AI can generate pretty much any format, it cannot generate journalistic storytelling. And it cannot make the human connections.

Implications for newsrooms: moving journalists up the value chain

AI and AI agents (software systems that use AI to complete tasks on behalf of users, or a way to automate a workflow by assigning an AI agent to each task) are already able to do a lot of the work in a newsroom.

In the near future, you could imagine each journalist being the editor-in-chief of a newsroom of AI agents, said David Caswell, founder of StoryFlow. Meaning the journalist would assign all routine tasks, including the writing of articles, to AI agents, while they themselves would dig up the good stories — the real value of journalism.

“We need a more story-centric way of working, more boots on the ground,” said Ezra Eeman, strategic advisor for WAN-IFRA’s AI in News Media Initiative. Gard Steiro, editor-in-chief and chief executive officer of VG in Norway put another way: “We need to move journalism and journalists up the value chain.”

Implications for users: increased relevance through personalisation

Not all consumers of journalism have the same needs. This is something many news publishers have worked more concertedly to address over the past few years.

With AI, the ability to deliver journalism based on individual needs has been turbo-charged. This leads back to the death of the article: With AI, an article can automatically be turned into another format depending on who the journalism is for and how they would like to consume it.

For example, a timeline for context, audio for people who struggle to read, or a vertical video for quick consumption. The journalism and data points are constant, but the delivery formats are liquid: Anything can be turned into anything else.

And all of this can be done without requiring time spent by journalists. Instead of editing videos, reporters can focus on finding the stories.

This has been the logic behind United Robots’ automated content services since long before generative AI came along. We use quality data and “old-school,” rule-based AI to produce unique local content, which adds value for users and saves time in the newsroom.

And we work with local news publishers, because, in local spaces, the time journalists spend out in the community is particularly impactful to the value they add to the journalism.

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