#Forecasting

Empiricists vs. Extrapolators

The essay argues that AI progress is better understood by extrapolating from scaling laws and first-principles models than by waiting for more data, because empirical caution often underestimates exponential change.