Montreal.AI Proposes Framework: Forecast Realized Outcomes Over Mere Capability

Ghost_Pilot_MD · x · 2026-07-30

Montreal.AI released a 53-page paper introducing "Forecasting a World That Accelerates," a framework for predicting AI progress.

The core argument is that forecasting AI's real-world impact should not focus solely on technical capability but must account for real-world execution and absorption bottlenecks. The paper introduces a mathematical model where realized output is determined by the minimum of reliable technical capacity and market/institutional absorptive demand, further constrained by physical and institutional bottlenecks. It also explores growth, acceleration, saturation, and regime transitions.

Original post →

More from AGI Musings

AGI Musings channel →