Let’s start with a disclaimer: E24 is owned by VG, Norway’s biggest tabloid, and is part of the Nordic giant Schibsted Media. This gives us an enormous advantage compared to similar-sized newsrooms and our main competitors because when AI experiments at Schibsted succeed, they are scaled and easily adopted by each news brand.

And, this is my first point: This sharing and scaling is not a given consequence of ownership and the organisation itself. You cannot pull a rabbit out of a top hat just because it is a top hat.

The magic at Schibsted is found in the fostering of a culture where people share and learn from each other. This culture is put into a system. This is the case with the AI task force, a group composed of representatives from each news media brand, where everyone learns from each other’s trials, failures, and successes.

Thanks to this, E24 has AI summaries, an AI buddy reader, an AI draft writer, and many other tools that are improving our journalism and making our workflow more efficient. So, if you are part of a big media group, you need to put in the work to make all parts of it coalesce into something bigger. It is with this sharing of experiences and knowledge that the magic happens.

I’m referring to the kind of magic, such as when E24 got a tip that Norway’s higher education minister had plagiarised her master thesis and one of our AI-literate reporters found the tool Copyleaks to conduct a document comparison using AI.

The news cycle took eight hours from our tip to when the minister resigned (the world’s fastest resignation?). AI literacy among newsroom staff proved crucial, and then we shared the method with VG, which later brought another minister down with the same approach.

To innovate or to be replaced

As a business publication, there are unique opportunities for E24 to grab. We need to drive innovation with AI within our niche. If we do not, somebody that does is likely to replace us.

E24 is a relatively small newsroom, with fewer than 50 employees. Most of them are on the editorial side, so we do have plenty of domain knowledge on business news. However, in many cases we are lacking enough technical expertise to run the AI projects and experiments we want to do, and, in many cases, feel like we need to do.

To remedy this, we have started to reach out. There are plenty of institutions, organisations, and businesses out there with similar motivations, and connecting has proven fruitful. Right now, we are involved in two partnership projects we believe will make an impact on our reporting.

Real Estate Analyzer

One of these projects is part of the JournalismAIFellowship, a collaborative programme with the aim of utilising AI to make journalism better. In a team of six, with editorial and technical resources from Gannett and Hearst in the United States and The Globe and Mail in Canada, we are working on an AI system to detect newsworthy real estate transactions.

E24 is part of the Journalism AI Fellowship Program, an initiative assembling almost 40 journalists and technologists from around the world to use AI to improve journalism.

Real estate stories are great at creating user engagement, and it is something that almost everyone relates to. Also, real estate is a way both to display and to hide power and wealth. We know we are missing out on stories on this beat; we believe automation and AI can help.

Price and well-known buyers and sellers are the most important signals. A solution showing promise is an automatically updated famous person list, which helps us match names from this list with names in the transactions data. When certain thresholds are reached, a large-language model (LLM) will create an alert to notify the newsroom.

Collaboration across borders also increases the exchange of different perspectives and experiences. If we are solving problems for several newsrooms, we can create systems that benefit journalism even more.

Improving our stock market coverage

Another project we are working on is a cooperation between E24 and the AI Journalism Resource Center at the Oslo Metropolitan University. At E24, we have started using GPT-4o and Claude 3.5 Sonnet to help draft news articles from stock company news notifications, with fairly good results at detecting the newsworthy information in the press releases.

But maybe we could use AI to get a better understanding of what notifications are moving the market?

Using advanced data analysis and LLMs, we see signs our approach can greatly benefit our stock market reporting. Academic data and AI expertise merged with the stock market reporter’s deep domain knowledge creates high-value results and progress at a higher speed than we could manage alone.

Competing together

Information systems are once again transforming. News gathering, production, distribution, and consumption will change. To adapt (and hopefully even to drive this change), we have to experiment, and we can do that faster and with more impact if we join forces.

Moving forward, we are not each other’s biggest competitors. We are competing to make objective, fact-based news relevant and competitive for the audience’s attention in a future-focused information ecosystem fueled by AI.

Cooperation makes it more likely we will succeed. Then we might pull a rabbit out of the hat.

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