The AI Journalism Toolkit We Actually Need

Newsletter sent 3.6.2025 by oler

In this issue: AI tools and prompts for journalists. A new 340-page trend report. The path to funny and creative LLMs. How Google’s AI Mode answers questions about news and elections. And solving the AI puzzle with SZ Institut’s Dirk von Gehlen.

What we’re talking about: Business Insider plans to lay off 21 percent of its staff. They’re selling this as a strategic response to AI search and the collapse of SEO traffic. Media companies are at a “crossroads” and can’t afford to waste time, writes CEO Barbara Peng in a company memo. The commerce business is supposed to take the biggest hit.

The clearer thematic focus – not just doing whatever because Google was driving traffic anyway – but actually focusing the brand, is supposed to be backed up by live events. That’s how owner Axel Springer wants to position Business Insider for a future where AI platforms control media distribution.

And then there is this: “we are going all-in on AI” to “work faster, smarter, and better.” You can obviously handle both things in one email, sure. But it’s also giving big “we’ll replace you with a chatbot” energy.

Media strategist Katharina Köth calls it “Axel Springer doing Axel Springer things,” to which media consultant David Caswell replies: “air cover from governance & shareholders probably makes the difference between acting on a strategy and meandering along doing business as usual.” Cool, cool.

Less than a year ago, Business Insider’s staff got a very different memo. It was a mail with book recommendations by a senior editor. Except, plot twist, some of the books didn’t exist, hinting at AI generation, reports Semafor.

If this feels like déjà vu, you know about the recent Chicago Sun-Times incident where a ChatGPT generated summer reading list was actually printed, including fake books. Or you might have heard it first at Berlin’s re:publica conference, where AI reporter Clare Spencer and I took stage. (Shameless plug: there is a YouTube recording if you’re into that sort of thing.)

What else is new:

AI & Journalism Links

LLM journalism tool advisor: Tired of figuring out how to use AI in journalism? Joe Amditis has collected tools and prompts and built a practical interface.

But AI isn’t creative, isn’t funny, isn’t surprising! Always just average, everything’s been done before. That’s kind of true – but here comes a paper saying: You can actually train LLMs to be creative too. It’s called Creative Preference Optimization.

The OpenAI-suing New York Times is entering into a multi-year partnership with Amazon, with content appearing on Alexa and going into AI training. (Jaspreet Singh, Reuters)

Good Tape transcription: How Denmark’s Zetland build an AI tool that brings in $3 million in annual recurring revenue. (Clare Spencer, Generative AI in the Newsroom)

Chatbot alert: Advance Digital and NJ.com launch AI-powered Politics Election Hub for New Jersey’s governor’s race 2025. (via David Cohn)

Back to the future: Does Axel Springer just see more clearly what’s coming for all of us? Let’s turn up our sensors! After 6 years in hiding, Venture Capitalist Mary Meeker is back with one of her famous trend reports. This time on AI. 340 pages of statistics.

With all due respect: there are slides in there that literally just consist of ChatGPT answers – namely: what AI can do well today, in 5 years, in 10 years. Still, the actual, not AI generated data tells a story.

The report talks a lot about CapEx, spending. That’s high. Revenue, on the other hand – well. And while the word “bubble” appears exactly once, when citing a book about railroad investments in 1840, the subtext is clear: we’re in that classic disruption cycle where everyone throws money at the shiny new thing.

The big takeaway: Unprecedented speed in the development, adoption and use of a new technology. So this isn’t just PR or some vague feeling: It’s happening very fast. It’s everywhere.

There’s a nod to the “next frontier” of AI, models that can operate flexibly across disciplines. Sounds good. But can we get these models to stop hallucinating fake book titles first? Maybe nail the basics before we start dreaming of utopia? And where’s the timeline for that?

But enough about numbers, let’s get to what really matters:

Three Questions with Dirk von Gehlen

One more thing: Google’s AI Mode is still just a beta version. It looks like Perplexity or ChatGPT – a big chat window, a few sample questions. And while Google search currently still shows links to publishers or headlines for queries about current events, AI mode delivers complete answers here too.

This even applies to election results, a topic where Google has been particularly cautious with AI Overviews and Gemini responses so far. This is obviously just a snapshot. In the US, AI mode is already being activated for all users, though still somewhat hidden.

That could change at any time, of course. More searches and screenshots in my blog post.

See you next time!

Subscribe to THEFUTURE

Get this newsletter in your inbox. For free.

The previous issue is AI Tools Gone Rogue, the next issue is The Generation That Gets AI.