Stories of the future are undermined by agent assumptions

According to LessWrong, most AI safety forecasts embed hidden assumptions about what future agents will be—humans, institutions, AI models, or hybrids—without ever testing those assumptions. These untestable beliefs shape everything downstream: what dangers we anticipate and what governance solutions we design. The author notes that traditional forecasting often misses discontinuities, like how new drone tactics unexpectedly reshaped the Ukraine conflict. Rather than chase doomed predictions, the piece advocates for designing antifragile systems—institutions that remain resilient even when our assumptions turn out wrong.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton