A London gettogether has focussed on opportunities for publishers as AI in media begins its move from a phase focused on efficiency and optimisation, to a revolution in the information ecosystem.

The Arc XP Connect event earlier this month included hosted a discussion with four AI experts, who outlined two distinct phases in AI’s trajectory.

Marketing head Dorinne Hoss says AI’s next phase “will likely transform the media space, including opening up new ways to share stories”. Her comments are published in an INMA Digital Strategies blog, and on the company’s website..

AI consultant David Caswell told how many organisations were deploying AI to enhance familiar tasks, from content tagging to audience targeting. “It’s a perfectly reasonable place to start; it makes sense to automate the stuff you already know, in order to optimise the status quo and make your life more efficient as it is.”

AWS principal solutions architect Mark Watkins shared a practical example in AI-powered article tagging: In the past, tagging thousands of articles manually could take months, but now, AI systems could tag content in days, significantly reducing costs and freeing up human resources for more strategic tasks, he said.

A standout example of efficiency through AI came from Venkat Varalu, vice president of data and analytics at the Washington Post, who described its “smart metering” system, which uses a deep-learning algorithm to optimise paywalls.

An AI system adapts dynamically, learning from real-time data to provide the best experience for each user. The approach allows the Post to compete in the crowded digital space, especially against big players such as Google, by using AI to level the playing field.

Hoss says that for journalists, “this efficiency phase is more than just a cost-saver; it’s a new tool for quality journalism.

Lars Adrian Giske, an investigative journalist at Mediahuset iTromsø in Norway, told the gathering how AI had sped up his investigative processes, cutting research time by as much as 80 per cent, by sifting through archives for newsworthy documents.

Giske emphasised that AI enabled journalists to focus on stories that might otherwise remain hidden, democratising investigative journalism by making it accessible to smaller newsrooms with limited resources.

Turning to the new information ecosystem, Caswell said the functionality AI was offering, “not just to the news industry but right across the information ecosystem”, added up to something “very, very different” to that of the existing ecosystem.

“We’ll go through a fairly radical transformation that may be as large as the transformation that social created. In my view, it will likely be larger.”

Giske said one of the most exciting developments in phase two would be AI-driven storytelling. “I think we’ll see totally new experiences,” he said. “For example, letting people explore stories in a different way than today, asking questions that maybe aren’t apparent in the story.

“We can use the data from an investigative project and have that available to the reader for them to explore within the context of that story space.”

Panelists also predicted hyper-personalised storytelling, where AI tailored the tone, format, and even length of articles based on the reader’s style and available time.

“Currently when we talk about personalisation, we are talking about moving modules around personalising a recommendation or what the next article is,” Varalu said. “But I think personalisation will go significantly deeper to changing the tone of the article depending on my style of reading, changing the length of the article, depending on how much time I have.

“That will include formats as well. If I prefer to read an article that has a video, there will automatically be videos created. These are all things we are talking about already.”

AI would also enable a new breed of journalism that spanned “vast information landscapes” with Caswell discussing the concept of “reasoning models”, such as OpenAI’s new tool, which can apply significant computational power during the task execution itself, rather than just during training.

These tools would allow for simple investigative journalism at scale, offering more detailed, precise outputs that can be trusted by both readers and journalists.

Members of the panel discussed the “endless” possibilities of phase two, but also the responsibilities. With AI’s growing influence, they agreed media companies would have to use it wisely, focusing on where it can make the most impact without forcing it where it’s unnecessary.

Varalu cautioned against overusing AI, suggesting that organisations concentrate on high-value tasks where AI can truly shine – such as automation of repetitive tasks or optimising digital content – rather than applying AI across the board.

And Giske emphasised that AI should support journalism’s “democratic mission”, helping to hold power accountable and inform the public.

For media companies embracing AI’s future phases thoughtfully and creatively, this technology holds the potential not only to increase efficiency but to create an entirely new, dynamic, and personalised way for people to experience news.

Watch the panel discussion here

Pictured (from left) Arc intelligence VP Joey Marburger (who moderated the event), with David Caswell, Venkat Varalu, Lars Giske, and Mark Watkins; (top) Lars Adrian Giske