A few days ago I was on a factory floor in Norway working on a process problem that would previously have taken several days of groundwork: data extracts, hypotheses, visualisations, code drafts and material for discussion with the operation. With today's AI tools, a first useful analytical foundation came together on the same working day. Not perfect. Not self-running. But fast enough to change how the work is organised.

That is why I say that AI is not just another digital tool. It is a system shift in how qualified work can be performed, reviewed and scaled. When the technology automates parts of analysis, formulation, document review, code support and other applied cognitive work, productivity is not the only thing that changes. Staffing, competence requirements, management and education also begin to move.[1]

Sweden has not lacked warning signals. The AI Commission delivered its roadmap earlier than planned, and the government has now also presented a national AI strategy. The problem is not total silence. The problem is that much of the practical response still happens at a normal pace on an area where capacity, costs and usage patterns can shift quickly.[2]

This is a proportionality issue

This is not an argument for panic. It is an argument for proportion. If AI had mainly been about better administrative tools, gradual implementation, general upskilling and a few pilot projects would have sufficed. But when the same technology starts affecting both private white-collar work, public administration, education, research support and industry-near analysis workflows, broader preparedness is needed.[3]

The public discussion often gets stuck between two bad extremes: either AI is treated as trendy software, or it is treated as science fiction. Both reactions miss the most important thing. What matters is not whether all the most far-reaching scenarios occur. What matters is that even the more grounded development is enough to create large adjustment requirements.

Three things the state should do now

First: establish a small but permanent national AI coordination unit with operational mandate across departmental boundaries. The AI question is too broad to be reduced to business policy or digitalisation. It affects the labour market, education, administration, total defence, research, energy and access to data.[4]

Second: shift focus from general ambition to prioritised use cases. Which parts of the public sector can benefit most from AI under human oversight? Which professions need transition support most quickly? Which data and procurement barriers can be removed in the short term?

Third: start speaking maturely about the labour market. It is possible to be a technology optimist and still realise that certain career steps, certain junior roles and certain administrative functions will come under pressure. If the state waits to discuss this until the effects show up in statistics, it will have waited too long.[5]

What is at stake

This ultimately concerns governance capacity. Countries that succeed in combining fast adoption with institutional adaptation will be stronger. Countries that get stuck in a mix of slow processes, unclear responsibility and symbolic initiatives risk losing both competitiveness and public trust.

AI is not everything. Energy, defence, demographics and geopolitics also play large roles. But AI amplifies several of these questions at the same time. That is why simply "having a strategy" is not enough. The question is whether the strategy actually changes priorities, mandates and implementation.

Source notes

The sources below support especially the labour market and policy arguments. The field observation in the opening draws on personal experience from industry-near improvement work.

  1. For how generative AI is used in tasks and professions, see Anthropic Economic Index, Economic Index report, January 2026 and PwC 2025 Global AI Jobs Barometer.
  2. AI Commission: The AI Commission's Roadmap for Sweden. Government: Sweden's AI Strategy, strategy document and action plan.
  3. Overviews of the breadth of impact: Stanford HAI, AI Index 2025, WEF Future of Jobs 2025 and ILO: Generative AI and jobs: A 2025 update.
  4. Swedish institutional capacity and strategic direction: Sweden's AI Strategy and AI Commission's roadmap.
  5. For labour market and productivity risks and opportunities, see IMF, WEF, PwC and ILO Working Paper 140.

Rolf Skogling writes AI-skiftet from an industry-near and practical perspective, grounded in working with AI in real operations.