AI for Data Journalism, New Formats and Personalization
In my series on AI in journalism, I’d now like to highlight some applications that are especially close to my heart: data-driven reporting, new storytelling formats, and personalized news experiences.
AI for data handling and analysis
AI-driven tools can find patterns and links in massive amounts of data. Connections that might otherwise not stand out to humans. These kinds of solutions have been applied to huge data leaks like the Implant Files back in 2019 – data journalists at SVT used machine learning to process the data – and the Luanda leaks in 2019.
As Quartz’s AI Studio explained in their article about the Luanda Leaks:
”ICIJ partnered with Quartz’s AI Studio to find a solution. We built a system using artificial intelligence to ’read’ all the documents and help journalists from Quartz, ICIJ and other partner organizations find the kinds of documents they expected in the cache of leaks—regardless of file format, spelling, transcription errors, phrasing, or even the language of the document.”
The increased use of AI has the potential to break many more data journalism stories. As put in the EBU News Report 2024:
”Data journalism is still in its infancy. Generative AI will help to scan and analyse huge amounts of data in a very short time, be it hidden in text documents, videos, or audio files. Humans will need to have the right research questions and working hypotheses. This potential should excite journalists.”
The EBU here points to the use of AI to give better access to unstructured data. Text documents, videos and audio files can be turned into data and skills in prompt engineering will be key for data journalists, as well as knowing how to form hypotheses about the data.
Tools like Google Pinpoint can help turn PDFs into structured data, saving time and making large document dumps more manageable. I’ve used it myself to sift through hundreds of accident reports. You still have to double-check and do corrections, but it saves time.
At the JournalismAI Festival, a collaborative project called Data Robot AIde was presented by teams from The Washington Post, Bloomberg, Il Sole 24 Ore, and India Today. This open framework uses LLM-powered chatbots to make extensive datasets — like the U.S. census with over 8 000 variables — more accessible to non-experts.
AI tools can also assist in the analysis, finding outliers or helping you with the programming code for doing so. You still need analytics skills though, since what you get back from the tools is not always correct. You have to be able to evaluate it.
The Financial Times have been experimenting with AI-powered tools to enhance their journalists’ ability to find insights in large datasets. In a Medium post about their efforts, they write:
”Our experiments have highlighted the transformative potential of automated robust data pipelines and AI-driven analysis in story finding. Data pipelines uncover hidden narratives within complex datasets, while AI equips journalists with advanced tools to perform more sophisticated analyses, elevating their story finding capabilities.”
AI for new formats
AI can also help reformat content for different platforms. For example, articles can be turned into short videos or podcasts using tools like Videofy, and synthetic voices can transform text into an audio news update. Several Schibsted outlets have tested synthetic voices for news podcasts, and the Brazilian investigative agency Agência Pública has done the same to reach broader audiences.
The text version of the story can also be written in different ways for different audiences, adapting to their needs, with the help of AI tools.
In an effort to engage younger audiences, Aftonbladet has even experimented with presenting news as rap music.
In a recent speech, Richard Fletcher, Reuters’ Director of Research, reasoned about these new opportunities for journalism.
”I am […] interested in how it might be possible to use AI to create additional versions and formats that are a better fit for how people access information in the digital world.
For example, taking articles written by journalists and using generative AI in the newsroom, or on news publishers’ own websites, to create summaries that convey the key points – or to convert that text to audio (or vice versa) to fit in with how people prefer to access media today.”
Chatting with the news can be a way to make information more accessible or interactive. Interacting with news via chatbots has become increasingly common. Schibsted’s Verdens Gang and Aftonbladet both tested chatbots for the 2024 US election, and The Washington Post has a ”Climate Answers” chatbot based on eight years of reporting from their climate team.
AI-powered news avatars. In early 2024, while studying the AI-driven journalism and media production course at Linneuniversitetet, I heard about Channel 1, where AI-generated news anchors were to deliver the news. The demo was impressive, and the launch was set to sometime in 2024. But now that I check their website it seems that while they have been posting some videos on X (the last one being from Sept 3), there has been no launch of a channel service. Rather, they seem to be offering AI-powered video services to other companies.
Modular journalism, breaking up the news.
– I still feel personally that there’s one field that is underdeveloped yet, and that is design or interaction interfaces for genAI content, said Alessandro Alviani, Generative AI Lead at Süddeutsche Zeitung Digitale Medien, in one of the JournalismAI Festival sessions.
– I think the chatbot interface is great, but still it can be limiting for a lot of people who basically don’t know what to type into it. And that’s why I think there’s huge potential in modular journalism. We have huge opportunities to start rethinking content and the way we structure and deliver content. If you combine that with personalisation, there is huge potential there.
In a lightning talk at the JournalismAI Festival, Priyanka Vazirani, Co-founder at Volv, presented the app they are building for creating bite-sized journalism for the TikTok age. Volv is a separate platform for journalistic content, presented in the shape of 9-second reads. The content is AI-driven and tailored to reader preferences. Each summary of a news story at Volv links back to the original article.
Personalization and localization
AI enables a deeper level of personalization in news, tailoring both content and format to individual user interests.
”Generative AI will help journalists to focus narratives, formats, and a tone of voice that engages audiences – ideally without compromising factual content and the message they want to convey.” – The EBU News Report 2024
”In a world saturated with content, personalisation is becoming paramount. Generative AI could craft experiences tailor-made for individuals.
Imagine a movie that adjusts its storyline based on your preferences or a video game that evolves based on your playing style.” – Sven Størmer Thaulow, EVP Chief Data and Technology Officer, Schibsted, in their Future Report 2024.
The EBU News Report also highlights new opportunities for hyperlocal news. Generative AI can create localized weather and traffic reports, and data-driven journalism can zoom in on specific regions. These stories, once too resource-intensive to produce, might now be feasible thanks to automated processes.
—
AI tools used for this blog post:
- Midjourney for the main illustration.
- Notebook LM for structuring the information, gathered from about 50 sources.
- ChatGPT for language corrections and improvements.
Previous posts in this series:
Recent Comments