I am sitting in a café with a young reporter who is frustrated.
Normally, I’d say she is a future star: an original thinker, strong voice, a clear sense of what journalism should do, and plenty of management ambition. She is, in other words, precisely the kind of person media organisations always claim they need. Now, but also, increasingly, in the future.
But right now, she isn’t the star. On the contrary, she can’t even get a job.
Why? Because she does not fit the template. Every time she meets with a hiring editor, she pitches ideas about community relationships, innovative formats, and audience value. And every time, the editor is impressed, knows that he or she should hire her, but has no place for her in their organization.
And also because the editor across the table—whom has many times been me—is simply trying to solve an immediate problem. The editor doesn't want to discuss long-term visions; he or she needs someone to cover the 6:00 AM breaking shift. Preferably yesterday.
The schedule had holes, and the holes always won.
Meanwhile, industry conferences and editorial retreats endlessly echo the same gospel: in the age of artificial intelligence, newsrooms must double down on what makes us uniquely human. We must create deep relationships, build communities, and attract the people who have a distinct voice. Like the reporter at the café. Still, we don’t.
This paradox has now reached a tipping point. The technology question in journalism is in many parts of the world more or less settled. Within a year or two, virtually every newsroom will have access to the same baseline AI capabilities. The gap between fast adopters and slow adopters in mature markets is rapidly closing.
When underlying technical capabilities become a commodity, "AI-driven product innovation" ceases to be a competitive moat. We are not there yet, and many news organizations – and markets – have a long way to go before they have adopted AI tools and, even further away, organized themselves around an AI mediated world. But that time will come. And as AI absorbs the mechanics of standard production, speed, and aggregation, the so-called "soft" qualities of journalism—editorial judgement, personal voice, and relational trust—stop being nice to haves. They become the core craft itself.
The next generation of news products will come from newsrooms that have radically reconstructed their organization, culture, and incentives around the things AI cannot replicate and which actually adds value in a completely new world and logic.
To understand what this looks like in practice, we have to look ahead. Let me take you to the newsroom I think we should be building, around six years from now!
The radically reconstructed newsroom
By 2032, the media landscape has drawn a sharp line in the sand, and most newsrooms have finally abandoned the dangerous middle ground.
On one side lies the automated information layer: fast, hyper-personalized, algorithmically synthesized "ambient news" that delivers commodity facts cleanly and instantly to users' devices (whatever they are in 2032). Publishers that manually tried to compete on speed and volume within this layer were, unsurprisingly, not successful.
On the other side are the human-centered editorial media.
Step inside one of these newsrooms on a Tuesday morning in 2032, and you will not find a traditional morning meeting organized around filling different sections on a homepage. You will probably not find separate, isolated "desks" for politics, local news, or crime either.
The newsroom could be organized entirely around audience need clusters: Help Me Understand, Give Me Perspective, and Connect Me to My Community.
Automated background systems monitor public records, transcribe council meetings, flag data anomalies, and maintain routine tracking feeds. The journalists do not touch this layer unless human verification is needed, or ethical questions are to be answered.
A new type of editor
Meet Maya, a Lead Curator & Perspective Editor (the same reporter from the café six years earlier, now leading a team).It is a silly title, but so was the behaviour in the newsrooms of 2026, where we asked ourselves “what should we publish today?” or “what’s the next article from the foreign news department?”.
2032 Maya does none of that. She reviews overnight signals synthesized by background intelligence tools and asks something like:
"What’s noise that we can ignore so our audience doesn't have to be bothered by it?"
"Where is human presence, moral clarity, or live investigation irreplaceable today?"
Her morning is not spent writing three rapid updates to keep an impression counter ticking. She spends two hours on the ground conducting an in-person investigation, followed by co-hosting a closed-loop audio forum with community members affected by a local policy shift. Later, she publishes a deeply voice-driven piece of analysis, evaluating why it matters and who should be held accountable.
The closed forums, the meetups, the Q&As where readers can talk back and journalists actually answer, that is the product, not just an engagement tactic bolted onto the bottom of an article.
Enabling new user behaviours
Take Leila, 54, a preschool teacher in a Stockholm suburb where the local school is about to be closed. In the ambient layer, she can get the decision, the budget figures and a neat summary in a matter of seconds. What she can't get there is someone who sat in the room, asked the politician the follow-up question she would have asked, and then, two weeks later, comes back to the forum and says: here is what has happened since we last talked. That is what she happily pays for in an editorial media product: the feeling that someone is paying attention on her behalf.
Maya is a well known figure in Leila’s community and people trust her. Her primary metrics are something we could call “Trust Density” and “Relational Dept”. Measures of subscriber retention, active user participation, and perceived value.Her 2032 newsroom operates on a simple premise (that probably is too obvious to be stated): AI handles the information supply; humans hold the relational trust that people pay for.
What it will take to get there
To make the 2032 newsroom possible, we have to stop treating AI as an efficiency tool designed to do old work slightly faster.
The promise of AI was always that it would hand time back to journalists. Yet, in most newsrooms today, those freed-up hours are quietly refilled with more output: more rewritten briefs, more social posts, more noise. We have automated the wrong half of the job.
If we want to build news products that endure, three foundational shifts must happen within editorial culture and skill-sets:
1. Stop measuring old values
If an organization’s incentives, KPIs, and editorial praise still flow toward publishing speed and raw click volume, nothing will change. Journalists will simply use AI to produce the same disposable content at double the velocity. Newsrooms must measure what actually creates long-term value: direct audience contribution, retention-per-interaction, and perceived trust.
2. Shift from Gathering Information to Editorial Triage
In an era of synthetic information overload, the primary value of a journalist is no longer gathering raw data; it is knowing what to ignore. Organisations like Fojo—and the broader journalism training ecosystem—were built on upskilling reporters for shifting realities. Today, that upskilling should mean teaching ruthless editorial triage: the capacity to slice through noise, evaluate algorithmic output, and exercise moral and factual judgment.
3. Cultivate voice and relational competence
Generative text is hyper-polished, perfectly balanced, and utterly bland. Human value lies in distinct perspective, lived experience, and intellectual honesty. Furthermore, journalists need skills in facilitation, active listening, and community leadership—turning passive consumers into active participants. As Swedish researcher Emma Frans noted in a different context: AI won't take your job, but someone more fun to be around might. I believe the same goes for products.
Translated to media: AI will not replace journalism, but newsrooms that build unique, authentic human relationships will easily outlast those that don't.
In pursuit of relationship-building journalists and profiles (“the voices” if you would), media leaders need to be mindful of the mismatch between future talents and legacy roles. If you recruit the most original, voice-driven talent in the country and slot them into a templated legacy role built for production output, you will most likely lose them within a year or two—to Substack, to podcast networks, or to platforms where they can actually build something of their own. Every time we pass on or underutilize an unconventional talent because they don't fit an existing shift, we risk funding our future competition.
So what should we do now?
Set KPI:s for trust and relationships. And make the editor-in-chief personally responsible for protecting it the day the 6:00 AM shift needs covering, because that day will come.
Every hour AI hands back to the newsroom should be accounted for. Decide in advance how much of it goes to relationships, presence and investigation, and write it down. If you don't, it will be eaten by more briefs and more social posts, just like today.
Remove one volume metric from the management dashboard and put a relational one in its place. Just one. Then watch what happens to the conversation in the morning meeting.
None of this requires waiting for 2032. It requires media leaders who, the next time someone like the reporter from the café walks in, stops asking whether she can cover the morning shift, and asks what she would build if we let her.