VRT explores AI-driven conversational news formats

At the heart of the initiative lies a key question: how can AI help make news more accessible and more personal, while remaining firmly anchored in a journalistic context?

Responding to evolving user behavior

The project builds on a clear trend. More and more people are turning to AI systems and chat interfaces to find information, including news. In Flanders, over a quarter of the audience already uses generative AI to search for news topics (imec.digimeter, 2026).

“This evolution encourages newsrooms to rethink how they present their content,” says Luk Overmeire, who initiated the research project.

Towards interaction-driven journalism

Traditional news formats—such as articles, videos and audio—remain essential, but offer limited opportunities for interaction. “These formats are important,” Overmeire explains, “but they do not easily allow users to ask follow-up questions or request additional context. Especially for complex topics such as geopolitics or major societal changes, that can be a barrier.”

Artificial intelligence can help address this challenge by searching large amounts of information and providing relevant answers. At the same time, it raises important questions. Many generative systems provide quick responses but do not always clearly indicate their sources, making it difficult to distinguish reliable information from misinformation. Established news media are also often underrepresented in these systems.

“At VRT, we are exploring a different approach,” says Kristin Van Damme, researcher at VRT Innovation. “We are developing AI applications that build on our own journalistic sources, editorial choices and storytelling. Beyond chatbots, we are experimenting with interactive interfaces that can tailor news to users’ needs.”

These interfaces could include formats that adapt to individual users or allow them to deepen their understanding of a topic over time, for example by saving and revisiting information. “AI is a tool to bring our journalism closer to the audience,” adds Overmeire. “Together with imec, we are also exploring how to make our news archive searchable for complex topics in the future.”

Human-centered approach

The research goes beyond theory. VRT has already tested conversational formats around major news themes, such as one year of Trump and the war in Ukraine. Instead of a single overview article, users can ask questions and receive answers from VRT NWS journalists.

This dialogue-based approach proves effective in helping users navigate complex stories. Those seeking more context receive additional explanations tailored to their needs. VRT is now investigating how AI can further support and scale this concept. A key principle remains that all answers are grounded in VRT’s own extensive and reliable news archive, ensuring that existing information is presented in a clear and accessible way.

Editorial oversight remains essential

A full chatbot offering is not expected in the short term. “If we make news available through chatbots, they must be fully reliable,” says Karel Degraeve, journalist and AI and innovation expert at VRT NWS. “Current AI systems still fall short in that respect, especially as our archive largely consists of video and audio, which remains challenging for these technologies.”

For that reason, editorial oversight remains crucial. VRT is currently focusing on interactive formats, such as ‘De Belastingspecial’, where AI is used to analyse the many questions from users and ensure that all relevant themes are addressed.

Trust as a key differentiator

The overall conclusion of the research is clear: the future of news is not fully automated, but human-centered and interactive.

“In a world overflowing with information, trust becomes increasingly important,” concludes Van Damme. “That is precisely where public service media like VRT can continue to make a difference.”