The MediaFutures Annual Meeting 2025 gathered 140 researchers, industry partners, and technology experts to discuss new developments in artificial intelligence for the media sector. This year’s theme, “Navigating Uncertainty with AI: Battling Misinformation and Empowering Users,” highlighted both the potential and the challenges AI poses for journalism, communication, and creative industries.
The meeting started with opening words by the Deputy Leader of the Infomedia Institute at UiB, Lars Nyre, Synnøve Kristine N. Bendixsen, Dean of the SV Faculty, Christian Birkeland, Chief Digital Officer at TV 2, and Christoph Trattner, Director of MediaFutures, cherishing the collaboration between academia and industry.
Expanding Capabilities and Creativity
The first keynote, by Cornelia Bjørke-Hill from Microsoft Norway, focused on using AI to support workflow and improve efficiency. She warned that human capacity has reached its limits: “Statistically, every second minute we are getting interrupted at work and have less time to focus, so we need support from AI. People often turn to AI because it’s fast, always available, and full of ideas.”
She shared practical examples from her daily work, including prompts and AI agents for tasks like social media posts and press releases. Her final advice was simple: “Whatever you use AI for, always do a final check before you use the content it creates.”
Following her, Alex Connock explored the creative possibilities AI opens for media professionals. He urged companies to rethink old habits before disruption forces them: “We must disrupt ourselves, because a big disruption is coming anyway.”
He emphasized that AI should be central to any media strategy: “The only value in your business is what the AI cannot do.”
Connock also highlighted emerging opportunities, from restoring old films to translating ancient scrolls, while cautioning against misuse and profiling. He predicted that by 2026, AI agents handling entire creative workflows would become mainstream.
Technology in Action
Speakers from Google, Microsoft, and Schibsted offered perspectives on AI in today’s media landscape.
Zhixian Bao from Google Norway highlighted the rise of multimodal AI: “Multimodality is a key trend in 2025. Forty percent of our time goes to finding information.”
Maxim Salnikov from Microsoft reminded the audience that humans must remain in control: “We humans control the agents. We set the direction. Success starts not with technology but with company culture.”
From a media perspective, Victorina Demirel from Schibsted noted how AI platforms are changing user habits: “People may no longer go to the original news source. We want to experiment responsibly, be transparent, and advocate for ethical clarity. I believe truth, trust, and technology can and must coexist.”
A deeper look at bias, trust and human signals
The next part of the programme shifted toward the risks, limits, and responsibilities that come with AI. Mor Naaman from Cornell Tech shared findings from a study where people used AI to write petitions on an online platform. The result was the opposite of what many hoped. The texts didn’t become better, they became “more similar and less diverse.” He said true empowerment comes from giving users “the opportunity to send the right signals.” When we rely too heavily on AI, we lose what he called the “lost signals” of communication: authenticity, ownership, effort, and attention.
From the Center for Digital Narrative, Jill Walker Rettberg warned about subtle cultural biases hidden inside AI-generated stories. She gave simple, memorable examples: “American stories are nostalgic, Norwegian stories are about fjords, German stories carry a subtle past trauma.” Her question to the room was direct: What happens when such underlying biases appear in speeches or media content?
Ingmar Weber from Universität des Saarlandes offered a different angle by asking what “slow” or “mindful AI” might look like. He reminded everyone that misinformation is not the only problem. “There is also a lack of tolerance and solidarity.” And not every person or every situation needs AI at all. As he put it: “Not everyone needs a car, right?”
Personalisation, newsroom tools and new routines
The following talks brought the focus back to practical solutions about how to strengthen trust, improve communication, and support journalists in a changing media landscape.
Daniel Catalán from UC3M presented both short- and long-term approaches, ranging from media and digital literacy in early education to better platform governance and community-based networks. He shared findings from a social experiment showing that “the most effective messages are narrative-based and positively framed.” Daniel also reminded the audience that uncertainty is a natural part of research: “Being 100 percent sure in science doesn’t exist and that’s okay.”
Next, Bart Goethals from Froomle spoke about the growing need for personalized news. “People expect personalized experiences,” he said, “but newsrooms must earn that trust.”
He demonstrated Froomle Voice, a multi-agent system where users can “chat” with the news as if having a private conversation. Bart explained how this type of interaction can support journalists by revealing which questions readers actually have, helping them write clearer and more relevant stories.
From ORF, Stefan Kollinger presented AI tools already in daily use in their newsroom. AiDitor transforms content into text, audio, or video, while the O-R-F Online Research function lets users “ask questions to ORF content” based on millions of archived pieces.
Stefan underlined one key lesson: “It’s all about the database.” And to maintain trust, he stressed the importance of being clear about which parts of the content are AI-generated.
Editors speak openly about real newsroom challenges
A panel of editors from DN Media Group, TV 2, Bergens Tidende, Amedia, and Schibsted discussed how their newsrooms are using AI and what challenges they face. While enthusiasm for AI was high, the conversation also highlighted practical and ethical challenges.
Chris Ronaldsen from TV 2 noted that discussions often focus more on AI benefits than difficulties, and Jan Stian Vold from BT observed that many journalists still struggle to use new tools effectively. Magnus Aabech stressed the importance of prioritizing user needs over trends, while Victorina Demirel and Erik Bonesvoll underscored responsible implementation, trust, and transparency.
Celebrating new talent
The day concluded by highlighting emerging researchers. Sixteen posters and four demos were presented, and both the scientific jury and the audience voted after one-minute pitches.
The winners were Yuki Onishi for her gaze-tracking project with Vizrt and TV 2, and Khadiga Seddik for her work on personalizing headline styles with LLMs.
The MediaFutures Annual Meeting 2025 demonstrated the centre’s continued commitment to developing responsible and human-centred AI solutions for the media sector. By combining academic research, technological innovation, and industry collaboration, MediaFutures aims to support a more informed, resilient, and trustworthy media ecosystem.
The centre thanks all speakers, partners, and participants for their contributions and looks forward to welcoming them again in 2026.
Centre Director Christoph Trattner:
“I am impressed by how much interest our MediaFutures Annual Meeting attracts, how professional we have become in organizing these meetings, and how many internationally recognized high-profile speakers choose to come and share their insights and knowledge on responsible AI — and do so entirely for free. Wow.”
