Stop Chasing AI. Start Solving Problems!

By now, everyone’s talking about AI. But in the media industry, where time is always tight and the pressure to innovate is high, the big question isn’t whether you should do something with AI — it’s what, why, and how. 

Aliya Itzkowitz has spent the last few years helping publishers across Europe make sense of it all. As a manager at FT Strategies, the consultancy arm of the Financial Times, she’s seen the AI wave up close from the first burst of excitement to the more sober realities of implementation. And if she’s learned one thing, it’s this: 

“Stop chasing AI. Start solving problems!”

From hype to structure

When generative AI first took off, many newsrooms rushed into action. “There was definitely a lot of pressure,” Itzkowitz says. “Everyone felt like they weren’t doing enough.” Some companies even made it a KPI just to use AI within the year, regardless of purpose. The result? A lot of experimentation, and a lot of confusion. To bring clarity, FT Strategies created the AI Launchpad, in partnership with the Google News Initiative. 

The goal wasn’t to push AI tools, but to help news organizations take a step back and look at their actual needs. “Forget AI for a second,” Itzkowitz says. “What are the key stages in your news business?” They boiled it down to three: news production, audience engagement, and monetization. Only once these areas were mapped did the question of AI come into play.

Build or buy? Try first.

Many publishers soon face a familiar dilemma: should we build something ourselves, or buy it from a vendor? There’s no one-size-fits-all answer. Buying usually means moving faster and with less risk. Building gives more control, especially when it comes to data and internal systems. But Itzkowitz points out something many overlook: building can be valuable simply because it helps teams learn. She recalls working with a financial newspaper in Bulgaria that built a summarization tool into their CMS. “The project itself was impressive,” she says, “but what really stood out was how much the team learned along the way.” That learning, she argues, can be just as important as the output. This idea, that the real value of building something is the skills and insight it develops, is what journalist Uli Köppen once dubbed the Trojan unicorn. It’s a phrase Itzkowitz loves. Still, you don’t have to start by building. Testing off-the-shelf tools can be a smart way to explore an idea and see how your team responds. “Trying before you buy is a very important part of the process,” she says.

When AI takes a detour

Experiments rarely go exactly as planned, and that’s often when they get interesting. One newsroom began testing a voice cloning tool so that journalists could record audio versions of their opinion pieces. But along the way, the project veered into something entirely different: helping improve a system that detects fake audio — part of a fact-checking initiative called VerificAudio.

“They were two separate projects,” Itzkowitz says, “but they found a way to connect the dots.” The FT’s own Ask FT chatbot also took unexpected turns. Built to answer professional readers’ questions, it soon found fans among internal staff who used it to search the archive. Some readers even used it to track specific journalists they followed, something the team hadn’t anticipated at all. In other words: fast, light experiments can uncover value you didn’t know to look for.

What matters now

After two years of pilots and prototypes, many newsrooms are shifting gears. The big question is no longer “Can we do this with AI?” but “Did it work, and is it worth scaling up?” FT Strategies now works with publishers to audit their AI activity, evaluate impact, and identify what should become part of the core business. In some cases, the most meaningful results aren’t in audience numbers or revenue, but in how a team has evolved when it comes to the way they think, collaborate, and explore new ideas. “Sometimes the best outcome of an experiment isn’t the tool,” Itzkowitz says. “It’s what it teaches you.”

Looking forward

While the AI landscape keeps moving, Itzkowitz believes the next step is not necessarily to do more, but to be more deliberate. That means asking tough questions: – What’s actually changing in the business model? – How are users engaging with journalism in new ways? – Are you building capabilities for the future, or just chasing a trend? There’s no clear roadmap yet, and no perfect playbook. But one thing is clear: the newsrooms that ask the right questions, and build the right learning culture, will be best prepared for what comes next. Because in the end, AI isn’t just about automation or efficiency. It’s about adaptation. And for that, the smartest strategy is to stay curious.

This article is based on Episode 6 of Kilkaya’s podcast, Insight Shapes News.

 Listen to our podcast