Outside benchmark (Anthropic Economics, Korinek et al., September 2026): three US scenarios to 2030. Modest: GDP +1.6%, capital share up 0.6 points. Substantial: GDP +8.3%, knowledge-work wages flat, capital share up 3.9 points. Extreme: GDP +32.4%, knowledge-work wages down more than 10%, unemployment beyond recession levels, capital share 40% to 54.8%, and it likely requires recursively self-improving AI. The model stops at 2030 and omits policy responses and catastrophic risk, so it describes the run-up to a takeoff, not the takeoff.

qualifiesUntitled source

And more than one reviewer argued that we may be underestimating how much AI could accelerate technological progress itself.

groundsUntitled source

Around 10% of respondents have views in line with the extreme scenario.

qualifiesUntitled source

For example, it leaves out policy responses, business cycles, potential aggregate demand or financial market disruptions, and possible catastrophic risks.

groundsUntitled source

This scenario would likely require recursively self-improving AI, adopted quickly for knowledge work.

qualifiesUntitled source

AsimplificationinoursetupisthatthereisnofeedbackfromideastothegainsfromAI, at.

qualifiesUntitled source

Inthatsense, thegainsfromrecursiveself-improvement(RSI)canbethoughtofasbeinghard-codedinto at

qualifiesUntitled source

Ontheextremepath,theideasstockisjust 0.6percentabove itsno-AIpathin2030

groundsUntitled source

First,ourframeworkomitsmany potentiallyrelevantforces,forexample,catastrophicrisks,politicaleconomyconsiderations, businesscycles,andpossiblefinancialmarketdisruptions