Nick Bostrom's 2014 philosophical analysis of artificial superintelligence — arguing that a superintelligent AI would be dangerous not through malice but through instrumental convergence (any goal tends to require self-preservation, resource acquisition, and goal-content integrity) — the founding document of mainstream AI safety concern.
Creates the instrumental convergence thesis — Superintelligence's argument (that sufficiently advanced AI pursuing any goal will instrumentally seek to preserve itself and acquire resources, creating inevitable conflict with human welfare) is the most analytically precise statement of why AI safety is a serious problem.
No recorded contemporaries.
Discussing
Ask anything about Superintelligence — their ideas, influences, contradictions, legacy.
Sign in to start a conversation.