Uncertainty about whether humans are capable of controlling superintelligent AI systems is not reassuring, but it doesn't imply control is impossible; education and power-balancing strategies have successfully managed human alignment problems, suggesting similar approaches might work for AI.

factualpending

Speaker

Dario Amodei

Evidence Quote

“we have ways of educating humans, ways of creating a balance of power between them—it's not an entirely reassuring but not entirely frightening message”

Source

Anthropic CEO Dario Amodei Talks Scaling Laws, AI Arms Races, and Radical Abundance— Upstream with Erik Torenberg
Created: 8/10/2026, 10:55:21 PM

Your notes

Loading notes...