Svea Eckert is a freelance tech journalist and expert for NDR/ARD television, Germany’s national public broadcaster. She will hold a workshop on “responsible reporting on AI” at the 2026 ESMH Summer School on science communication in the age of artificial intelligence. We asked her a few questions about what good AI coverage should look like and what participants can expect from her workshop.
How did you first get into tech journalism, and what drew you to covering artificial intelligence (AI)?
Svea Eckert: My interest in technology journalism began about 15 years ago. It started with the Google cars showing up in the streets in Germany, taking pictures of every house. It was both so interesting and so disturbing. Around the same time, Facebook introduced the “Like” button, which sparked controversy in Germany, and disputes over copyright and monetisation emerged between YouTube and GEMA, the German rights management agency for artists. The new economy from the Silicon Valley was starting to grow and that’s when I decided that accurate tech coverage was needed. The Internet is not a faraway country, it’s right here with us and it is impacting our society.
I was an economy reporter before, but then I decided I wanted to cover tech and never regretted it. It’s getting more exciting every day because the field is evolving so rapidly and its impact on our society is even stronger now with the rise of generative AI.
You are currently working on a project titled ‘Between Hype and Doom: Why Good Journalism on AI (Still) Falls Short’, supported by the Stiftung Mercator Fellowship Program. What questions are you trying to answer, and what have you discovered so far?
Svea Eckert: After 15 years of fast-paced tech reporting, I now have the opportunity to do some research in my own field for a year, analysing in a scientific way how we, as tech journalists, report on AI. Looking closely, you often see a pattern: AI coverage oscillates between breathless enthusiasm and apocalyptic warning – neither of which is actually helping the public to understand what’s going on.
So, the core questions I’m trying to answer are: what gets in the way of nuanced, accurate AI journalism? Structural constraint or editorial pressure? Is it a lack of expertise? Is it time? I mostly conduct interviews with journalists across German newsrooms, and what I see is a real tension between the speed of the news cycle and the time it takes to genuinely understand the technology. This often affects the depth of reporting. How we can improve our reporting on AI is something I want to develop during the fellowship.
And what about public perception? In your view, do audiences generally perceive AI with more optimism or concern?
Svea Eckert: The conversation is often very abstract, between “AI will change everything” and “AI is dangerous”. I see that often, there’s not enough grounding: which are the systems and who are they used by? Who is doing what with whose data? Who is accountable to whom? These questions don’t get asked enough. Coverage is improving on the economic and labour dimensions, especially on jobs. But for me, what matters the most is a human-centred approach, which I think is mine too: what is actually changing for people? How is AI impacting us as humans and as a society?
What do you think is still missing from the public discourse on AI?
Svea Eckert: What is still missing, in my view, is stronger reporting on the power structures behind AI – this is something I’ll stress at the ESMH Summer School. The market is extremely concentrated, from chip production to software and AI models, and controlled by a small number of very powerful companies, and mostly men. Keeping this in mind helps avoid falling into the hype-or-doom narrative, which often originates from the companies themselves. As journalists, our role is to look behind the curtain and report on what is actually happening and how it affects people and society.
Does accurate AI coverage require specialised science journalists, or can generalist reporters handle it too? And how well-equipped are most newsrooms right now?
Svea Eckert: Ideally, journalism needs both expertise and the ability to explain complexity to a broad audience. A specialist who cannot communicate clearly is not very effective in a mainstream newsroom, but neither is a generalist who does not fully understand the technology they are covering. Right now, many newsrooms are still struggling or improvising. There are pockets of excellence, but they are not the norm.
Education helps to get into the right direction, but I also think we cannot stay siloed within science journalism, political journalism, economic journalism and so on. It is important to stay in close contact with each other, exchange knowledge, and understand what each of us is doing. This is something I am trying to do right now, connecting German tech journalists from different media to work together and share experiences.
Last month, a former Italian politician and founder of the Democratic Party (PD) published an interview with Anthropic’s Claude in the Italian newspaper Corriere della Sera. He asked it about its childhood, desires, and even its relationship to God and death. It sparked quite a debate. What do you think of this kind of journalistic experiment?
Svea Eckert: These systems can produce remarkable responses, which is both fascinating and revealing of what we project onto machines. But I personally think it’s a journalistic trap. When you publish it as an interview, you must frame it clearly because you are implicitly treating the system as a subject with inner experience, and that’s a claim that requires extraordinary evidence. What bothers me isn’t the experiment itself: it’s the lack of interrogation around it. What matters to me is not what the system says about topics such as death, but why it says it. What is the model actually doing? Who built it, and for what purpose? I think this is journalism and the rest is theatre.
You have already given us a glimpse of it, but what else should participants expect from your workshop at the 2026 ESMH Summer School?
Svea Eckert: The workshop is built around a simple premise: how do you cover AI without being fooled by itself or by the people selling it? We will work through concrete cases, look at how to evaluate claims made by the companies and researchers, and we will talk about specific pitfalls. It will be hands-on and interactive, because I want people to leave with something they can actually use the next morning. I will emphasise the human-centred approach so that they can open the curtain, and get a sense of how to engage with AI topics without falling for hype or doom narratives.