Published 11 hours ago • loading... • Updated 14 hours ago
Will AI really kill everyone? How, exactly?
Researchers say the most plausible risks still face major physical and technical barriers, while critics call the extinction claims sparse and unfalsifiable.
Last week, Anthropic employee Jacob Coxon quit, citing dangerous work, sparking alarm among politicians and news organizations. Fellow Anthropic worker Evan Hubinger estimated the chance of AI-powered human extinction within 10 years at greater than 10%.
Speculation among doomsayers includes AI-driven bioweapons and autonomous killer robots. Researcher Thomas Larsen, from the AI Futures Project, detailed these scenarios in reports titled "AI 2027" and "AI 2040," projecting different developmental and regulatory frameworks.
Nate Soares, president of the Machine Intelligence Research Institute, authored "If Anyone Builds It, Everyone Dies: Why Superhuman AI Would Kill Us All." Cheng Xing countered that appealing to superintelligence is "handwaving" without measurable physical evidence.
Heidy Khlaaf, chief AI scientist at the AI Now Institute, argued that "Scientific claims require falsifiability precisely to avoid the nature of religious arguments." She noted nuclear facilities are "air-gapped" from the internet, limiting AI access.
Herbert Lin, a senior research scholar at Stanford University and member of the Science and Security Board at the Bulletin of Atomic Scientists, said AI amplifies risks but the material threat remains with nuclear weapons, which have existed for more than 80 years.