In this episode of “Aimpactful,” we explore the intersection of artificial intelligence, journalism, and community building with Mattia Peretti. As an ICFJ Knight Fellow and an experienced Learning Experience Designer, Project Manager, and AI Consultant, Mattia has played an important role in guiding news and media organizations toward adopting AI technologies.

Mattia Peretti‘s journey into the field of journalism and AI began in 2019 with the JournalismAI project at the London School of Economics. Fostering AI literacy within the journalism community is something for which he is now widely recognized.

His collaborative efforts include helping The Guardian create AI guidelines, facilitating AI transparency discussions for Swedish publishers, and aiding Internews in establishing an internal AI community of practice.

In this video podcast, Mattia shares his insights on how journalists can embark on their AI journey, stressing the importance of understanding the technology, its limitations, and its ethical implications.

We also speak about his ICFJ Knight Fellowship, which is dedicated to bringing together a coalition of people who believe that AI technologies can indeed help the journalism industry, but only if we first collectively accept that our mission is not to be content creators but to provide a service built on listening.

And on that note, we warmly invite you to listen to or watch “Aimpactful” right now.

Transcript of the AImpactful Vodcast

Branislava Lovre: Welcome to Aimpactful. Today, we’ll speak about learning and using AI in newsrooms. Our guest is Mattia Peretti. Welcome, Mattia.

Mattia Peretti: Thank you for the invite, Branislava. Very happy to be here and have this conversation.

Branislava Lovre: We’ll talk about your career and important projects, but let’s start from the beginning. How did you first get involved in AI and journalism?

Mattia Peretti: The answer to this question is always very overwhelming, to be honest, because it was a bit of a coincidence. My relationship with AI started back in 2019 when I was invited to join the JournalismAI project during its early stages at the London School of Economics. Until then, I didn’t have a specific interest in or background in AI or computational journalism. I was working at the European Journalism Center in the Netherlands, focusing on journalism innovation, organizing trainings, events, and networking opportunities. I was essentially a project manager.

When JournalismAI was about to be created by Professor Charlie Beckett at the London School of Economics, in partnership with the Google News Initiative, it was meant to be a small, one-year research project, so they needed just one person to manage it and make it happen. At the time, I was leaving my previous job, so the timing was perfect. I was offered this opportunity and had no idea it would become my career focus for the next six years. It was a very fortunate coincidence, and it gave me the opportunity to learn a lot and meet loads of smart people over the years, especially before the generative AI wave made us all pay attention to this topic.

Branislava Lovre: You prepared different courses regarding AI literacy, and I attended one. It was a great experience, but what was so important is the community you built.

Mattia Peretti: Yes. Well, I’m very glad, first of all, to hear about your experience taking the course and that you still remember it as a positive experience. We’re talking about JournalismAI Discovery, which in a way is the last thing I did at JournalismAI before leaving the project a bit more than a year ago. I’m really proud of that course because it’s something that I had been thinking of creating for a while.

During my years at JournalismAI, I had the opportunity to design and create training materials on the topic for journalists. For example, mini courses on the Google News Initiative Training Center back in 2020, which introduced the concept of machine learning and its use cases. There was also the training program that became The JournalismAI Academy for Small Newsrooms, which is still ongoing. With Discovery, we really wanted to scale the value of these initiatives by bringing together four years of project knowledge into bite-sized content experiences delivered through email.

So JournalismAI Discovery was an email course. I don’t even remember if the modules were six or eight, but it was designed so that people and journalists could receive the content comfortably in their inbox and consume it at their own pace. Importantly, it also built a community of people taking the course together. In the first cohort, there were already hundreds, if not thousands, of journalists taking the course simultaneously. This meant it wasn’t just an individual experience but a shared journey where participants could interact with each other.

What was special for me was seeing the conversations happening in the Slack channel and how global it was. Hundreds of journalists from every continent participated, from around 70 countries, I think. This is crucial because AI has wide-ranging impacts on societies, and these impacts vary in different parts of the world due to cultural differences, language barriers, and bias. Going through this experience together and learning from each other about their priorities and innovations in different regions has been one of the most fascinating insights from JournalismAI.

So yes, Discovery is something that I am very proud of, and I believe it’s still ongoing. It was a team effort. I brought the idea from previous experiences and other courses that I had the opportunity to take. But it was really a great way for me to leave JournalismAI, with my little baby, as I used to call JournalismAI Discovery.

Branislava Lovre: If journalists want to learn about AI, where should they start? What should be their first steps?

Mattia Peretti: That’s a very challenging question because the answer could be many different things. I believe it needs to be a conscious effort that starts with various elements simultaneously. It’s important for everyone working in journalism not necessarily to become experts in using AI, but to understand what these technologies are capable of, their limitations, the risks of using them in certain ways, and the opportunities they present.

This understanding is crucial. But I also think it’s important to just try things out, especially now that generative AI is so accessible. We can all open our browsers and try out tools like Gemini, ChatGPT, or Perplexity. This gives us direct experience of the potential and the excitement of using an interface that provides answers. However, we must also be acutely aware of how these answers are produced and the potential errors in them, as we cannot trust them blindly.

As an industry, journalists still have a lot of work to do to better understand AI so that we can help society understand it. Many people still believe that ChatGPT simply looks up answers on the internet and provides accurate responses, which is not how it works. This misunderstanding can lead to dangerous situations if not addressed.

The possibilities for using AI are endless, but we always emphasize its responsible usage. AI has incredible potential and significant risks and limitations simultaneously, and we need to understand how it can truly help us. This means going beyond the obvious uses, which is a personal frustration of mine.

In journalism, even after more than a year, it feels like we’re mostly going for the low-hanging fruit—using generative AI to make our work more efficient, more productive, and to churn out more content. While these are not inherently bad things, they are not enough. We are missing an opportunity if we don’t spend more time figuring out how to use these technologies to genuinely deliver on our missions—informing societies better, engaging more users, and stopping the trend of news avoidance.

The issue isn’t that people are avoiding us; it’s that we’re failing to provide them with relevant content that they want to engage with. I genuinely believe that AI brings us opportunities if we are aware of them and if we are strategic and clear about our focus. If we continue only looking for efficiencies, it will keep us going for a little longer, but it won’t have any fundamental impact on our industry, which has been in crisis for 15 to 20 years.

Branislava Lovre: When and why did you decide to choose that topic?

Mattia Peretti: I spent the first six weeks of my fellowship reading and studying a lot. As trivial as it sounds, I went through various resources like the reports from the Writers Institute, which are released every year and talk about our news consumption and how our creation processes evolve over time. I also talked with people who are already making significant contributions to the industry and are headed in the right direction.

I’m in the process of understanding what good can come if we bring together these people and what they’re doing, trying to multiply the value of their work and make it bigger than the sum of its parts. Overall, stepping back and understanding the industry’s problems is something I’ve been building on over my ten-year career. There are many data points from those ten years that contribute to finding a solution.

Fortunately, I didn’t wake up on January 1st this year and suddenly realize there was a problem. It has been a long process of digesting what’s good and bad about our industry and trying to figure it out. In a way, it’s about what I can contribute with my skills and expertise, and honestly, what I’m passionate about. I strongly believe we do a much better job when we’re working on something we’re genuinely excited about. So there’s an element of selfishness in using this fellowship opportunity to deliver value to the industry, but also to myself.

Branislava Lovre: Recently, you wrote an article emphasizing “why” before “how” when we talk about AI implementation. You invited colleagues to reach out if the topic resonated with them. Were you satisfied with the response?

Mattia Peretti: Yeah, exactly. And I’ve been very humbled by the responses. I got so many that I stopped counting them—emails, comments, and messages on LinkedIn. I feel like I biased the process a bit by asking people to get in touch if the topic resonated with them, so I wasn’t necessarily inviting criticism. Maybe I should have also added a line asking those who disagreed to get in touch as well.

A lot of people shared the frustrations I highlighted and the conviction that focusing more on the users and listening to them more could benefit the industry. I’m not saying anything new; it’s not like I came up with a revolutionary idea. There’s already a lot of great work being done on user needs that many news organizations have embraced. Yet, there seems to be something missing. There are still barriers making it harder to turn this knowledge into action or strategic action, despite many good initiatives. We often revert to the old habits of how journalism has been done for decades, and I think we need to find a way to break that cycle.

I received positive feedback, but also polite responses saying, “This is all true, but we already know this.” I agree, but it still feels like we’re not doing enough in that direction. Maybe this is an area we need to explore further. I strongly believe that this is not something one of us can figure out alone in their office. That’s why I so strongly believe in collaboration and bringing people together. By combining the sentiments and expertise of many who are already doing smart work in this industry, we can make a push.

I’m currently designing a couple of meetings with a small group of people, with the help of a fantastic designer facilitator. She introduced me to a metaphor of a flywheel—a big heavy wheel that you want to start rolling. Initially, it’s incredibly hard to make any change; you’re pushing and nothing is happening. But slowly, you start making painful progress. As it begins to move, continuing becomes easier because the hard part was the initial push. That’s where I want to focus this year: figuring out how to keep this push going so it sustains itself, rather than just being another moment where we all agree on the need for change but the wheel doesn’t actually move.

Branislava Lovre: You work a lot on connecting people. Your idea was to create a directory of AI consultants and trainers worldwide with ICFJ and JournalismAI.

Mattia Peretti: At the beginning of the year, I was receiving loads of requests from both news organizations and media development organizations. They were asking if I was available to support their training programs or help with AI implementation in the newsroom. There were lots of interesting opportunities, but with my fellowship commitments, I had to say no to many of them. Every time I declined, I tried to help them find someone else who could assist them.

So, I shared on LinkedIn that I was thinking of creating a list of other consultants or trainers who could help organizations when they reached out to me. Clearly, I was onto something because dozens of people, including yourself, replied very kindly, expressing interest and availability to help. This led me to think, “Okay, the idea is good, let’s create this list.” But it was getting bigger than I thought, so I decided not to do it alone if I wanted to have time for my fellowship.

I reached out to my former colleagues at JournalismAI, as they seemed to be the best partners to make this happen. They were very kind and enthusiastic about collaborating. With their support, we first put together a form for anyone interested to fill in their details. Through that form, we created a public directory, which is now available on the JournalismAI website. If it hasn’t changed, it should still be at journalismAI.info, so check it out there along with all their incredible resources.

The directory now has, as far as I know, about 80 or 90 names of consultants and trainers from all across the world, with different types of expertise. They can help news organizations or anyone else in the media space with their AI needs, whether it’s training for AI literacy within their team or assisting with specific implementation challenges.

Branislava Lovre: Everything is changing day by day, so we have to learn a lot. Could you give us general advice in the journalism field?

Mattia Peretti: I think the key is to approach AI not as the solution or something with inherent value, but rather in the context of why you want to work in journalism in the first place. Take a step back and consider that working in journalism today does not necessarily mean being a journalist in the traditional sense. Inside a newsroom, editorial is just one department. There are also engineers, developers, and product managers. We need more of them because a product mindset is critically important. User research, audience experience, and design all contribute to the mission of journalism in a fundamental way that is as valuable as writing articles or appearing on camera.

My main suggestion for anyone wanting to work in journalism, like you and me, who believe that journalism is important and has a role in society, is to avoid thinking that being a journalist only means becoming a good writer. Working in journalism can mean many different things. Newsrooms need a variety of profiles, and even more if you can bring a mix of skills. It’s not about being the best expert at one thing, but about bringing a combination of skills.

We went through a phase where we talked a lot about the importance of bridge roles in news organizations—people who can act as a bridge between engineers, design teams, and editorial colleagues. It’s sometimes hard for these different departments to speak the same language and collaborate effectively. You need people who can bridge these gaps and take everyone along the journey.

So that would be my suggestion. AI is a part of it, but I don’t think my recommendation for anyone trying to enter journalism right now is to become an AI expert. Yes, do your homework. Understand AI, try it out if you’re passionate about it and want to do more, which is amazing. But it’s not a requirement that I would recommend anyone focus too much on.

Branislava Lovre: We already covered this, but the last question for today: what will shape the future of AI?

Mattia Peretti: I feel like the most important aspect is to really try to get creative about this, and not repeat the same mistakes or try to solve the same problems in the same way. We need to be much more collaborative and look closely at existing solutions and what people have experimented with, even if they didn’t build a final solution.

I was looking back at the projects developed during the JournalismAI Collab in 2020, which was our first experiment in getting people from different organizations to work together on a project. That program has now evolved into the JournalismAI fellowship, which is in its fifth year. The challenges they were working on included how to use AI-powered summaries to better inform readers, how to make the most of our archives, and how to use machine learning to build loyalty among readers. These are the same problems we’re trying to solve today, which highlights two important elements: first, we’re not going to be the final solution for any of this; second, technologies are evolving. In 2020, generative AI didn’t exist, so they were looking at AI technologies that we don’t focus on as much anymore because we’re all focused on generative AI right now. Yet, good solutions were already available and built at that time.

It’s about doing our homework and building on what others have done rather than starting from scratch and wasting time. We need to find new things while acknowledging the importance of the current issues. Let’s collaborate more, talk more. There are a lot of amazing resources—not only on the JournalismAI website, but many more outlets like podcasts, newsletters, and conversations like this one. I know we’re all busy in journalism with little time to do our homework, but doing it will save us time in the long run and be beneficial.

I wish there was a way to build the final repository of all AI journalism knowledge and make it easily accessible to everyone. That might be utopic, but until we get there, we should utilize the available resources and keep communicating with each other.

Branislava Lovre: This is the perfect message to end the interview. Thank you so much for your time.

Mattia Peretti: Of course. Thank you.

Branislava Lovre: You watched another episode of Aimpactful. Thank you, and see you next week.