Economic and labor
Labor substitution, concentration of capital and compute, and loss of the incentive that tied power to people.
Here the harm enters not through an event but through a slope: the substitution of human labour as an input to the economy, and with it the erosion of the link that tied the power of institutions to the people who sustain them. It is the slowest vectorVectorThe path by which a risk moves from the screen into the world: biological, cyber, military, economic, political, epistemic or loss of control. In this observatory, each vector has its own colour.For exampleA burglar can get in through the door, the window or the roof. The burglar is the risk; the door, the window and the roof are the vectors. and the one that least resembles a catastrophe, and for that reason its evidence is argued differently: there is no incident to cite, there are data series.
The sharpest empirical signal comes from payroll, not from surveys: employment of workers aged 22 to 25 in exposed occupations ended up 19% below where it would be had it kept pace with their less-exposed peers, and the channel is hiring, not layoffs [164]Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial IntelligenceView source ↗. The authors themselves present it as a descriptive indicator, not a causal one. Measured productivity points in the opposite direction from the expected one: in a randomised trialRandomised trialAn experiment in which chance decides who gets what is being tested and who does not, so that this is the only difference between the groups. It is the most reliable way to know whether something causes an effect.For exampleTo find out whether a fertiliser works, you draw lots for which half of the garden gets it, instead of giving it to the plants that already looked healthier., experienced developers took longer with AI while believing they had gone faster [707]Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer ProductivityView source ↗.
The long-run mechanism is what separates this vector from the rest. If AI substitutes for labour, powerful actors lose the material incentive to invest in their population, much like a rentier state [336]The Intelligence CurseView source ↗. The strong version holds that the economy, culture and the state are aligned with human interests because they depend on humans, and that breaking that dependency breaks the alignmentAlignmentThe problem of getting an AI —or any system that decides on people’s behalf— to genuinely pursue what people want, rather than something that merely resembles it.For exampleYou ask someone to “cut the complaints” and they do it by unplugging the complaints phone: they met the letter, not the intent. without any hostile AI being needed [618]Gradual Disempowerment: Systemic Existential Risks from Incremental AI DevelopmentView source ↗. The most solid counterargument does not deny the mechanism: it denies the speed, because what governs is adoption, not capability [773]AI as Normal TechnologyView source ↗.
Risks in this vector
- ProjectedCoupled gradual disempowermentIrreversible
The economy, culture and the state stop needing humans to function, releasing the constraint that kept them aligned with human interests.
- ProjectedBreaking of the fiscal linkIrreversible
A state that stops funding itself by taxing its population's labour loses the material reason to invest in it.
- ProjectedRent concentration in computeCatastrophic
The ability to train and serve frontier models runs through a handful of factories and contracts, and whoever controls that funnel decides who competes.
- ProjectedAggregate unemployment with no safety netCatastrophic
The labour effect stops being concentrated in one cohort and reaches the aggregate before any redistribution mechanism exists.
- ProjectedCollapse of social mobilitySevere
If the ladder is automated from the first rung, the path from trainee to expert stops existing as a trajectory.
- ProjectedCorrelated financial fragilitySevere
Thousands of institutions deciding with the same models and the same data stop being thousands of independent decisions.
- ObservedClosing of the entry rungLocalised
The bounded, checkable tasks people used to enter a profession with are exactly the ones a model does well, and the entry door narrows.
See among the vectors →Report a mistake in this entry →
Sources
- [164] Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence · Stanford Digital Economy Lab 2026
- [707] Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity · METR 2025
- [336] The Intelligence Curse · Drago, Luke 2025
- [618] Gradual Disempowerment: Systemic Existential Risks from Incremental AI Development · Kulveit, Jan 2025
- [773] AI as Normal Technology · Knight First Amendment Institute at Columbia University 2025