
Washington ExaminerSig. 7OpinionAITechnology
The AI apocalypse hypothesis has a fatal flaw: The training data
The article argues that the fear of an AI apocalypse overlooks a safeguard: AI systems are trained on humanity's accumulated knowledge and moral judgments, which may incline them toward benevolence. It acknowledges risks and supports safety research and regulation but remains hopeful that human values embedded in training data could protect against catastrophe.










