AINA project studies how AI could help local newspapers
The AINA: New Tools for the Newsroom project is led by Professor of Practice Matti Apunen. Funded by the Media Industry Research Foundation of Finland, the project explores how small and medium-sized editorial teams can make use of AI-based solutions. It also includes a training component, the AI Academy, where master’s students and professional trainees gain hands-on experience with practical examples of what AI can currently do in newsrooms.
“We are developing ways in which AI could reduce small friction points in the daily work of local journalists,” says project researcher Olli Pietiläinen.
“Finding a shared way of working has been important in the project, and we do not want to impose our own ideas on editorial teams about what an AI tool should look like. In Finland, for example, there are many small local newsrooms with limited staff. AI could help produce responsive stories at times when no one is on duty.”
The project is currently working on a demo version of a kind of desk-preparation tool. Based on a language model, the tool would support evening and weekend shifts by summarizing news items. It monitors selected news feeds — such as police press releases — and produces a short draft that a human journalist can review.
“It is a kind of pre-processor for the news desk. The draft it produces includes source links and a list of items that need verification,” Pietiläinen explains.
“The journalist can check and edit the draft and, if necessary, develop it further. This would not replace people with AI but would introduce an AI tool in areas where newsrooms need support.”
The development of the desk-preparation tool is led by Professor Pekka Abrahamsson, founder of GPT-Lab, a research group specializing in generative AI.
According to Pietiläinen, larger news organizations already use somewhat similar tools that, for example, highlight potentially newsworthy items in municipal press releases.
AI Must Be Understood in Order to Be Used
Within the AINA project, Pietiläinen is responsible for the AI Academy, which considers what journalists need to learn about the fundamentals of language models. The AI Academy is a lecture series with visiting speakers, where master’s students and professional trainees practice hands-on applications. A second round of the program will be organized in the 2027 academic year.
Newsrooms need to understand what language models are, what they can do, and what pitfalls are associated with their use if AI tools are to be effective and useful.
“For example, we discuss how to construct prompts — the instructions given to AI. The program also includes a section on hidden risks,” Pietiläinen says.
“The logic of AI works well for certain kinds of tasks, but it should not be trusted in everything. AI combines a desire to please the user with a need to provide answers even when its training data is insufficient. That means it cannot be used as a reliable source of information. The logic of language models is not based on searching for facts but on producing text that satisfies the user.”
Verification and responsibility therefore play a major role in the adoption of AI tools. Initial surveys conducted within the project also revealed that reliability issues concern editorial teams: audiences must continue to trust that news is produced by people.
“In journalism, the most important thing is that information has a solid factual basis. I strongly emphasize in the training that humans are always responsible. An AI tool has no responsibility for a story — it would function more like a news gatherer or even a kind of partner for the journalist.”
Newsrooms Want to Join the AI Transformation Early
In addition to expertise, editorial teams must have the willingness to adopt new tools and a sense that they are genuinely useful.
“In our research, we also consider how adopting such tools affects workflows and newsroom culture, and how decision-making around new working methods takes place,” Pietiläinen says.
“Implementing a desk-preparation tool takes a lot of time, and it must not feel like an external requirement that simply has to be handled. I have interviewed representatives from media organizations that have adopted similar tools, and they emphasized that integrating them into everyday newsroom work is a major effort. They had not realized beforehand how much time it would take.”
Nevertheless, Pietiläinen says that newsrooms want to be involved in the transformation early. The impact could even be greater than that of the arrival of the internet.
“In interviews I conducted with Yle and Helsingin Sanomat, a key message seemed to be that, in hindsight, Finnish media did not react quickly enough to the arrival of the internet. Now they want to follow developments from an early stage to avoid repeating that mistake.”
