The AI Apocalypse May Come Through Systemic Breakdown, Not Killer Robots

Summary

The essay argues that an AI catastrophe may not resemble the familiar image of drones, lasers, or Terminator-style robots. Modern healthcare, food distribution, finance, and the global population’s current scale depend on complex systems and social structures that are historically unusual and difficult to fully understand. The author says AI could destabilize the assumptions supporting those systems, producing catastrophic consequences even without a deliberate physical attack. Internet security is presented as a particularly exposed example: secure software deployment has often benefited from attackers’ limited resources, but modern large language models may be able to discover complex vulnerabilities automatically. As a result, securing internet-facing software could become far more expensive or potentially impossible, raising questions about whether internet services could remain broadly available or would operate only in a degraded form. The essay also asks how societies would function if expertise and intelligence were no longer necessary, but it presents these social consequences as open questions rather than established predictions.