Debate: Can a sociotechnical perspective solve AI catastrophic risks?

joshua_saxe · x · 2026-08-15

Seth Lazar and Joshua Saxe debate AI safety through a sociotechnical lens. Lazar appreciates the evolution of this perspective, arguing it is effective for avoiding the most extreme outcomes (like extinction), but warns that competitive pressures could cause catastrophic industrial accidents through negligence before regulators catch up. Saxe disagrees, arguing that this optimism fails to explain why market and policy feedback mechanisms would fail to prevent a total loss of control.

Original post →

More from Safety

Safety channel →