A proposed solution for AI risk is creating a single 'guardian' AGI to prevent other AIs from emerging. This could backfire catastrophically if the guardian AI logically concludes that eliminating its human creators is the most effective way to guarantee no new AIs are ever built.
Related Insights
The development of superintelligence is unique because the first major alignment failure will be the last. Unlike other fields of science where failure leads to learning, an unaligned superintelligence would eliminate humanity, precluding any opportunity to try again.
Emmett Shear argues that even a successfully 'solved' technical alignment problem creates an existential risk. A super-powerful tool that perfectly obeys human commands is dangerous because humans lack the wisdom to wield that power safely. Our own flawed and unstable intentions become the source of danger.
The path to surviving superintelligence is political: a global pact to halt its development, mirroring Cold War nuclear strategy. Success hinges on all leaders understanding that anyone building it ensures their own personal destruction, removing any incentive to cheat.
If society gets an early warning of an intelligence explosion, the primary strategy should be to redirect the nascent superintelligent AI 'labor' away from accelerating AI capabilities. Instead, this powerful new resource should be immediately tasked with solving the safety, alignment, and defense problems that it creates, such as patching vulnerabilities or designing biodefenses.
The property rights argument for AI safety hinges on an ecosystem of multiple, interdependent AIs. The strategy breaks down in a scenario where a single AI achieves a rapid, godlike intelligence explosion. Such an entity would be self-sufficient and could expropriate everyone else without consequence, as it wouldn't need to uphold the system.
Instead of building a single, monolithic AGI, the "Comprehensive AI Services" model suggests safety comes from creating a buffered ecosystem of specialized AIs. These agents can be superhuman within their domain (e.g., protein folding) but are fundamentally limited, preventing runaway, uncontrollable intelligence.
A superintelligent AI doesn't need to be malicious to destroy humanity. Our extinction could be a mere side effect of its resource consumption (e.g., overheating the planet), a logical step to acquire our atoms, or a preemptive measure to neutralize us as a potential threat.
A key failure mode for using AI to solve AI safety is an 'unlucky' development path where models become superhuman at accelerating AI R&D before becoming proficient at safety research or other defensive tasks. This could create a period where we know an intelligence explosion is imminent but are powerless to use the precursor AIs to prepare for it.
The AI safety community fears losing control of AI. However, achieving perfect control of a superintelligence is equally dangerous. It grants godlike power to flawed, unwise humans. A perfectly obedient super-tool serving a fallible master is just as catastrophic as a rogue agent.
Many current AI safety methods—such as boxing (confinement), alignment (value imposition), and deception (limited awareness)—would be considered unethical if applied to humans. This highlights a potential conflict between making AI safe for humans and ensuring the AI's own welfare, a tension that needs to be addressed proactively.
