Finland aims to run its entire public sector on AI by 2031 — but a leading researcher calls the plan unrealistic
Permanent Secretary at Finland’s Ministry of Finance, says the state, municipalities and regional health authorities will move to an AI-based operating model within five years, targeting a 20% productivity gain. A University of Helsinki AI researcher has since challenged both the figure and the feasibility of the plan.
Text by Martti Asikainen, 6.7.2026 | Photo by Adobe Stock Photos
Finland’s entire public sector is to run on artificial intelligence by 2031, Juha Majanen, Permanent Secretary at the Ministry of Finance, told Helsingin Sanomat, Finland’s largest daily newspaper, in an interview published on Friday.
Majanen said the goal covers central government, municipalities, and the wellbeing services counties — the regional authorities created in a 2023 reform to run healthcare and social care in Finland.
Finland’s public sector currently employs around 700,000 people.
A shared platform, and fewer jobs
The ministry’s plan centres on a common AI platform for the entire public sector, hosted on Finnish cloud infrastructure and continuously updated to run whichever commercial AI model is currently the most capable.
The aim, Majanen said, is to protect funding for Finland’s welfare state as the population ages and economic growth stagnates.
AI would take over tasks such as reviewing lengthy documents — work currently carried out by specialists, Majanen said. He was blunt about the consequences for staff: most reductions would come through attrition, as retiring workers simply aren’t replaced, but he acknowledged that many public sector employees will lose their jobs outright.
“AI will, in practice, replace people,” he said.
Where the 20% figure comes from
The Ministry of Finance says the shift would lift public sector productivity by at least 20% by 2031. But Jarkko Levasma, the ministry’s ICT director and the official leading the AI project, later admitted no formal calculation underpins that number — it is a best-case estimate. The ministry plans to publish three separate scenarios of AI’s likely effects this autumn.
Anna-Kaisa Ikonen, Finland’s Minister of Local Government and Regional Affairs, nonetheless called the target realistic when asked. The ministry maintains the changes should come at no net cost to the state and without compromising citizens’ legal protections or language rights.
Finland has two official languages, Finnish and Swedish.
Questions over feasibility
Teemu Roos, a University of Helsinki computer science professor specialising in AI, told the Finnish news agency STT that Majanen’s claims raise serious questions.
“This is, in my view, a very eyebrow-raising claim — what exactly might such a calculation be based on?” Roos said of the 20% target, though he stopped short of offering his own estimate.
Roos was also sceptical of Majanen’s core proposal, arguing that leading AI companies are unlikely to allow their most capable models to run inside Finnish-controlled infrastructure, since those models are commercial trade secrets that firms are reluctant to expose to outside scrutiny.
Persuading major US AI companies to operate this way will not be straightforward, he said, though he pointed to Microsoft’s arrangement offering access to OpenAI’s models under contracts that keep data within Europe as one possible model.
A civil servant speaking like a politician
Roos also questioned the tone of Majanen’s remarks — specifically the claim that Finland has no alternative but to adopt AI as widely and quickly as possible. That framing, Roos said, sounded more like language from an elected politician than a civil servant.
“I’m used to hearing this from politicians, that there are no alternatives, but it strikes me as strange that a civil servant would put it this way,” he said. “There are always alternatives, and these things should always be open to discussion.”
Majanen, for his part, admitted some unease with the pace of the technology, saying he had realised over the winter that AI could already match human experts on certain tasks. “I thought we were in a nightmare — close to the singularity, a point beyond which it is no longer possible to see,” he told Helsingin Sanomat.
The government’s economic policy committee has already begun organising the transition, setting up a cross-departmental leadership group and a transformation office to coordinate ministries, agencies, municipalities, and wellbeing services counties.
What this could mean for the region
For municipalities, wellbeing services counties and public-facing organisations in our region, the practical questions will centre on procurement, data governance, and workforce transition timelines — none of which have yet been detailed publicly.
Organisations working with or advising public bodies on AI adoption may want to watch for the ministry’s promised autumn publication of its three productivity scenarios, which should offer a clearer picture than the current headline figure.
There is also a broader lesson in how this initiative has been communicated so far. A headline productivity figure was released well ahead of the underlying methodology, and it took a researcher’s public challenge, rather than the ministry’s own disclosure, to surface the fact that no formal calculation exists.
Regional organisations planning their own AI adoption may want to take note: publishing ambition before evidence can invite exactly this kind of scrutiny, and a more cautious sequencing, evidence first, headline figures second, may serve smaller organisations’ credibility better than it appears to have served the ministry’s.
