Breaking of the fiscal link

The argument is not about technology but about the incentives of those in charge. A state that funds its spending by taxing the labour of its population has a material —not moral— reason for that population to be healthy, educated and productive: it is its tax base. A rentier state, which draws its income from a resource, does not. Luke Drago and Rudolf Laine transfer the resource-curse literature to AI: with AI capable of substituting for labour, powerful actors lose the incentive to invest in ordinary people ^[336]The Intelligence CurseDrago, Luke; Laine, Rudolf · 2025 · reportView source ↗Accessed on 9 September 2026. They open with the Democratic Republic of the Congo: more than 24 trillion dollars in untapped minerals and 73.5% of the population below 2.15 dollars a day in 2024.

The empirical step the argument needs is for income to shift from labour to capital. The International AI Safety Report 2026 puts it in the conditional: AI adoption may shift income from labour towards owners of capital ^[538]International AI Safety Report 2026Bengio, Yoshua · 2026 · reportView source ↗Accessed on 9 September 2026.

What this does not demonstrate. The resource curse is not deterministic, and the authors themselves cite the evidence that contradicts them: with comparable per capita rents and long-standing autocrats, infant mortality in Oman fell from 159 per thousand in 1971 to 9 in 2010, while in Equatorial Guinea it went from 263 to 109 over the same period ^[336]The Intelligence CurseDrago, Luke; Laine, Rudolf · 2025 · reportView source ↗Accessed on 9 September 2026. Pyramid-shaped replacement from the bottom upwards is an extrapolated prediction, and the fact that its first rung coincides with youth hiring data does not validate the whole chain. And the series that would have to break has not broken: labour’s share of US GDP has stayed around 60% for more than a century ^[618]Gradual Disempowerment: Systemic Existential Risks from Incremental AI DevelopmentKulveit, Jan; Douglas, Raymond; Ammann, Nora et al. · 2025 · paperView source ↗Accessed on 9 September 2026. In Denmark the effect on earnings and hours is a precise null that rules out effects larger than 2% ^[535]Still Waters, Rapid Currents: Early Labor Market Transformation under Generative AIHumlum, Anders; Vestergaard, Emilie · 2026 · preprintView source ↗Accessed on 9 September 2026.