Sam Harris — Can We Build AI Without Losing Control?

Sam Harris 2016 TED Talk remains the clearest articulation of why AI alignment is a serious philosophical and technical problem — not science fiction. Harris central argument: if we build a superintelligent system and its goals are not precisely aligned with human values, the result is catastrophic regardless of whether the system is malevolent. The comparison to a nuclear reactor without a shutdown mechanism and to a corporation optimizing for profit without ethical constraints are frameworks that have proven more useful than most subsequent AI safety communication. The remarkable thing about this talk in 2026 is how accurate its predictions were: Harris anticipated that AI progress would be rapid, that society would be unprepared for it, and that the difficulty of specifying human values formally would be the core technical problem. The talk is 17 minutes and has been viewed 8 million times. It is the best starting point for understanding why AI risk is a serious field of inquiry.