N/ANight Ash
Back to collection
Research

AI 2040: Separate Forecasts from Observations

Different assumptions require distinct scenarios, not one confidence score.

Published Updated
A gravel trail divides into two branches around a low rock outcrop in green mountains.
Technical illustration

The AI 2040 debate contrasts expectations of extraordinary progress with a slower model of capability and social change. When the starting assumptions differ that sharply, a single confidence score hides more than it reveals. Observed facts, model assumptions and investment decisions should remain separate. A forecast may influence preparation without being promoted into observation; retaining the evidence set and model version makes later changes in judgment explainable.

References