There is a period ahead of us — perhaps five years, perhaps longer — that is likely to be rougher and more unstable than much of what we have become used to during the first phases of digitisation. Not because collapse is predetermined, but because several shifts can coincide: faster AI development, uneven labour market transition, geopolitical tensions and already strained public systems.
This essay is therefore not an exact forecast. It is a scenario attempt: a way of thinking in waves and indicators to plan better.
What is happening right now
Spring 2026 makes it clear that AI is already affecting parts of knowledge work for real. Not in every profession and not in every organisation, but broadly enough that leaders, unions, education systems and public agencies should treat the question as structural rather than experimental.[1]
What makes the situation special is not just the capacity in the models, but the combination of better performance, easier access and falling usage thresholds. When tools become both better and cheaper at the same time, adoption can go faster than in previous digitisation waves.
Three waves to plan for
Wave 1: cognitive automation. Analysis, report writing, document review, simple code, administration, customer dialogue and other screen-based tasks change first. Here jobs rarely disappear all at once; instead work content, staffing and competence requirements shift rapidly.
Wave 2: agentic workflows. The next step is systems that can hold together more steps in a process under human supervision: fetch information, compile material, suggest decisions, coordinate between systems and keep track of multiple sub-goals. This can change both business structure and public administration.
Wave 3: more physical automation. The third wave concerns the combination of AI, machine vision, robotics and specialised systems in logistics, manufacturing and other more physical environments. The time horizon here is more uncertain than in pure cognitive automation, but the direction is clear enough to be a planning question already now.[2]
Signals to follow
If one wants to avoid both alarmism and naivety, one should follow indicators rather than slogans. Some of the most important are: how quickly real workflows go from pilot to everyday use, how many professions get clearly changed work content, how well agentic systems work under supervision, how quickly costs fall for sufficiently good performance and how early unions, education and public sector begin to adapt.
Another important signal is the information environment. When synthetic content becomes cheaper to produce and harder to verify, pressure on public debate increases just when society needs more shared orientation, not less.
Why the Nordic region still has better preconditions than many others
The Nordic region does not have a perfect model, but we have several institutional assets: relatively high baseline trust, established systems for transition and adult education, habit of handling labour market questions jointly and societies that in international comparison are digitally mature.[3]
That does not mean everything is well. Care pressure, skill shortages, fragmented politics and financial constraints are real problems. But compared to many alternatives, there is still a foundation to build on.
What must happen before turbulence feels inevitable
We need a labour market dialogue that treats AI as a structural force, not as another digitisation project. We need faster and more modular education. We need better analysis of which tasks change first, so support can come earlier.
We also need leadership that dares to speak about uncertainty without falling into either denial or doomsday language. The worst combination is when large changes come gradually but still faster than institutions can respond to.
Five years can be a very short time
I have seen several technology shifts up close: lean, automation, digitisation. What distinguishes the AI shift is the combination of breadth, speed and how close the technology comes to the actual work content. Therefore I believe that the next five years can be decisive — not because the world ends otherwise, but because institutions that do not begin to adapt now risk falling behind in a way that becomes expensive to repair later.
Turbulence is not the same as catastrophe. But whoever wants to avoid catastrophe must begin to prepare whilst development is still possible to influence.
Source notes
The essay is a scenario attempt. The sources support the labour market and capacity background, not exact dates.
- WEF Future of Jobs 2025, ILO 2025 update, Anthropic Economic Index and PwC 2025 Global AI Jobs Barometer.
- For AGI/agent perspective and security questions, see Google DeepMind, RAND and International AI Safety Report 2026.
- For Swedish and European digital context, see Sweden's AI Strategy and the EU's Sweden 2025 Digital Decade Country Report.