Top AI Researchers Predict 10x Productivity Gains in 2 Years, Economists Still Forecast Only 4-5% GDP Growth
soumitrashukla9 · x · 2026-09-18
Dwarkesh Patel's new podcast brings together John Schulman, Ollie O'Neill and Beren Millidge to debate RSI, Chinese labs' progress, automated AI researchers, long-horizon RL and why RL works. Economist Benjamin Moll highlights the bullish consensus: 10x productivity uplift for AI researchers within 2 years, and superhuman performance on all computer work in 5-10 years (Schulman: 3-4 years). Yet citing a Forecasting Research Institute working paper surveying economists, AI experts and superforecasters, Moll notes 10-15-year US GDP growth forecasts remain just 4-5% or lower — even among those who believe AI could dwarf the industrial revolution.
More from AGI Musings
- Are there three scaling axes left: training, reasoning inference, parallel agents? — BlackHC · 2026-09-18
- Doom rhetoric makes it harder to hold AI companies accountable, say researchers — sarahbmyers · 2026-09-18
- It's a literacy crisis: agents are dice rolls in production, not minds — gerardsans · 2026-09-18
- Hinton tells Congress 'maybe a year' to regulate AI, but critics say his doomer calls were all wrong — firstadopter · 2026-09-18
- Gary Marcus resurfaces 2023 Senate warning: we rushed into agents without sandboxing — GaryMarcus · 2026-09-18
- AI governance is about real institutions and power, not just extinction risk — Dr_Atoosa · 2026-09-18