What happens to human wages in an AGI economy?

This theoretical model introduces a distinction between bottleneck work (tasks essential for economic growth — output cannot expand without them) and accessory work (non-essential tasks — output can expand even if they are discarded). The model shows that all bottleneck work is eventually automated, while some accessory work may remain human.

The wage mechanism is the key contribution. Before AGI, "wages reflected the importance of bottleneck work and the scarcity of labor with the requisite skill for this work." With AGI, wages reflect something entirely different: "the computational cost of replicating the work produced by all human labor." Human labor retains value not because it produces something irreplaceable, but because it saves scarce computational resources. Workers are paid the compute cost of their replacement, not the economic value of what they produce.

This has a direct connection to the tokenization thesis. If intelligence is tokenized — valued by reception not substance — then the AGI wage model describes the exchange rate between human intelligence-tokens and compute-tokens. The exchange rate converges to parity: a human hour of work is worth exactly the compute it would take to replicate that hour. Since Does incremental AI replacement erode human influence over society?, the wage-convergence mechanism is the economic expression of disempowerment: humans remain employed but their labor is valued as a compute substitute, not as a human contribution.

The model also shows that "the share of labor income in GDP converges to zero" — even though people still hold jobs. Output shifts from being multiplicative in compute and human effort to being additive. Long-run growth is pinned by compute expansion, not labor force growth. Extending to scientific work: "without AGI for science, technological progress is constrained by population growth. With AGI, all scientific bottleneck work is automated and the rate of technological progress is determined by the growth rate of compute." This generates sustained exponential growth despite shrinking population, but does not create a singularity.

The accessory work residual is significant for the Tokenization of Intelligence framework: human labor survives in tasks where "it is impractical to use compute since we already have too many workers." Human work persists not from irreplaceability but from compute-allocation efficiency — the same logic by which cash persists alongside digital payment systems. The human contribution becomes accessory, supplementary, a rounding error in the growth equation.

Source: Work Application Use Cases