Kunstig intelligens gir neppe en ny, industriell revolusjon

Tobias Lunde · article · 2026-04-10 Reader 🖥️

AI Bubble Debate Technology Diffusion & Transformation Curves

History shows that AI is unlikely to produce an economic revolution comparable to the Industrial Revolution, because AI does not remove fundamental physical constraints on production the way steam and electricity did, because technology diffusion takes decades, and because broad value creation depends on institutional and infrastructural preconditions that are not yet in place.

theoretical · argumentative — An opinion piece (kronikk) using historical analogy to argue against the prevailing narrative that AI will produce an industrial-revolution-scale economic transformation.

Structure

Part	Role
Opening claim	States the thesis: history gives us good reasons to expect AI's effect on economic growth will be smaller and slower than many assume.
The prevailing narrative	Presents the common framing: experts and strategists compare AI's potential to the Industrial Revolution, positioning AI as a solution to falling productivity, labor shortages, and aging populations.
Historical corrective on the Industrial Revolution	Argues the common view misunderstands the Industrial Revolution — economic growth and urbanization began in the 1600s, well before steam power. Technology was a necessary but not sufficient factor.
Nature-of-constraint argument	Steam and electricity removed fundamental physical constraints on production (energy source, energy distribution). AI operates within existing physical and energy frameworks and does not remove analogous constraints.
Diffusion lag evidence	Historical examples (steam engines limited to mine pumping for 50+ years; ARPANET to public internet took ~25 years) show that decades pass before technologies produce broad economic growth.
IT productivity paradox	Despite information technology dominating modern life, productivity growth in developed economies has been declining since the 1970s. Even optimistic interpretations find little trace of the expected gains.
Systemic preconditions	Technological revolutions depend on politics, institutions, infrastructure, energy, competence, property rights, financial systems, and organization of work. Norway specifically lacks key preconditions (dependence on US cloud/platforms).
Closing	Reframes the question: not just what AI's potential is, but what is needed for broad value creation and whether those preconditions are in place.

Key terms

industriell revolusjon (Industrial Revolution) Technology Diffusion & Transformation Curves
Not merely the invention of steam power, but a sustained increase in the rate of change in economy and society, beginning in 17th-century Britain — driven by a confluence of factors, not technology alone.
fysiske begrensninger (physical constraints)
Fundamental limits on production imposed by reliance on rivers, wind, and muscle power for energy — constraints that steam and electricity removed but AI does not.
produktivitetsvekst (productivity growth) Technology Diffusion & Transformation Curves
Measurable increases in economic output per unit of input — the metric by which technology's economic impact should be judged.

Propositions

explicitAI will have a smaller and slower effect on economic growth than many assume.conf: high
explicitThe comparison between AI and the Industrial Revolution rests on a misunderstanding of what caused the Industrial Revolution.conf: high
explicitEconomic growth, urbanization, and the shift from agriculture to production began in Britain in the 1600s, before steam power.conf: medium
explicitSteam and electricity removed fundamental physical constraints on production; AI does not remove analogous constraints.conf: high
explicitcorroboratedIt typically takes several decades before technological advances result in broad economic growth.conf: high
supports ←Ordinary Engineers, Not Heroic Inventors · p8 — Confirms the 'decades before broad economic growth' pattern and supplies the organizational (not technical) explanation for why the lag happens.
explicitDespite IT and the internet dominating modern life, productivity growth in developed economies has been declining since the 1970s.conf: high
explicitNorway lacks key preconditions for extracting large economic gains from AI, particularly due to dependence on American cloud solutions and software platforms.conf: medium
implicitThe AI-as-revolution narrative is used to legitimize large costs and a fear of being left behind.conf: high
implicitAI is more analogous to IT/internet than to steam/electricity in terms of its economic impact pattern.conf: medium

Arguments

AI will not produce an Industrial Revolution-scale economic transformation. strength: medium

Explicit premises:

The Industrial Revolution was not primarily caused by technology (steam power) alone — economic growth and structural change preceded steam by over a century.
Steam and electricity removed fundamental physical constraints on production; AI operates within existing physical and energy frameworks.
Technology diffusion takes decades before economic effects materialize.
IT/internet — the most recent major technological shift — has not produced the expected productivity growth despite decades of adoption.

Implicit premises:

The relevant measure of a 'revolution' is sustained, broad-based economic growth and productivity increase.
If the most comparable recent technology (IT/internet) failed to deliver expected productivity growth, AI — which is a subset/extension of IT — is unlikely to do better.
Physical-constraint removal is a categorically more powerful economic force than information-processing improvement.

Inferences:

Since the Industrial Revolution was caused by more than technology, comparing AI to the Industrial Revolution based on technology alone is flawed.
Since AI does not remove physical production constraints the way steam/electricity did, it cannot produce the same kind of transformation.
Since even IT failed to produce expected productivity gains, expecting AI to do so is unwarranted.

Norway specifically is poorly positioned to capture AI economic gains. strength: medium

Explicit premises:

Technological revolutions depend on institutional, infrastructural, and political preconditions.
Norway depends on American cloud solutions and software platforms for data storage and processing.
Central parts of the value chain and infrastructure lie outside Norwegian economy and political control.

Implicit premises:

Countries that do not control the core infrastructure of a technology cannot capture its primary economic value.
Norway's existing economic model (resource-based) does not naturally translate into AI-driven value creation.

Inferences:

Norway's structural dependency means AI gains will accrue primarily to the US platform providers, not to Norway.

Narratives

Steam engines were limited to pumping water out of mines for over half a century before they became good enough to power factory machinery, and it took additional decades before factories were redesigned to fully exploit them.
→ Illustrates that even transformative technologies take a very long time to produce economic impact — supports the diffusion-lag argument.
ARPANET was established in 1968 but the internet did not become publicly available until the 1990s.
→ A more recent parallel reinforcing the diffusion-lag argument — even within living memory, transformative tech took 25+ years.
Factories with giant steam engines driving many machines were reorganized around assembly lines with many small electric motors.
→ Illustrates the nature of physical-constraint removal — electricity separated production and consumption of energy, enabling fundamentally new production organization.

Questions addressed

✅ Will AI produce an economic revolution comparable to the Industrial Revolution?
No — history gives strong reasons to expect smaller and slower effects.
✅ What actually caused the Industrial Revolution?
A confluence of factors beginning in the 1600s, not technology (steam) alone. Technology removed physical constraints but was one factor among many.
✅ Is Norway well-positioned to benefit economically from AI?
Not obviously — dependence on US platforms and lack of control over core infrastructure are significant barriers.
⭕ What would be needed for AI to produce broad economic value?
Raised but not answered in detail — the author notes it requires changes in infrastructure, laws, institutions, but does not specify what these would look like.

Gaps

incompleteThe argument treats AI as operating purely 'within existing physical frameworks,' but does not consider whether AI might remove cognitive/informational constraints that are analogous to the physical constraints steam removed. If cognitive bottlenecks are the binding constraint on modern economies (as many economists argue), then AI could be categorically transformative in a way the author's framework does not account for.
Read next: Read Daron Acemoglu & Simon Johnson 'Power and Progress' (2023) for a nuanced treatment; also Erik Brynjolfsson's work on the 'productivity J-curve' which argues IT productivity gains are real but delayed by complementary organizational change.
uninformedThe article cites 'nyere forskning' (recent research) on pre-steam economic growth without naming sources, making the historical claim difficult to verify. The revisionist historiography (e.g., de Vries's 'Industrious Revolution,' Broadberry et al.) is more nuanced than presented here — some scholars argue the pre-1750 growth was qualitatively different from post-Industrial Revolution growth.
Read next: Jan de Vries, 'The Industrious Revolution' (2008); Stephen Broadberry et al., 'British Economic Growth, 1270-1870' (2015).
incompleteThe IT productivity paradox is presented as settled evidence, but the debate is ongoing. Robert Gordon, Chad Syverson, and others have offered multiple explanations (measurement error, diffusion lags, winner-take-all dynamics). The author does not engage with the possibility that IT gains were real but captured narrowly.
Read next: Robert Gordon, 'The Rise and Fall of American Growth' (2016); Chad Syverson, 'Challenges to Mismeasurement Explanations for the US Productivity Slowdown' (2017).
closed byOrdinary Engineers, Not Heroic Inventors — Paul David's 'Dynamo and the Computer' account (cited here in full) supplies one of the concrete diffusion-lag mechanisms the gap says is missing from the productivity-paradox debate: electrification's payoff was delayed because factories kept old plant layouts, not because the technology itself was weak.
illogicalThe Norway-specific argument about US platform dependency is presented as an AI problem, but the author himself notes it is 'et problem uavhengig av KI' (a problem independent of AI). This weakens its use as evidence specifically against AI-driven growth in Norway.
Read next: Examine EU digital sovereignty strategies and whether platform dependency has actually prevented European countries from capturing IT-era productivity gains.