Earlier this week, former Google CEO Eric Schmidt, along with Dan Hendrycks, director of the Center for AI Safety, and Alexandr Wang, CEO of Scale AI, published a paper arguing against a government-led Manhattan Project to develop AI superintelligence and achieve national AI supremacy.
The authors say that if the US pursues an AI-dominance strategy, other countries (notably China) will do the same. They warn this will lead to unintended consequences and perhaps even war—be that a short-lived, inadvertent “flash war” that pits one nation’s AI against another’s, or World War III and omnicide, in which all human life is snuffed out.
Instead of racing toward dominance, they argue, the US should engage in a three-prong strategy of deterrence, nonproliferation, and competitiveness, which will lead to a global stasis, which they label Mutually Assured AI Malfunction (MAIM), inspired by the one brought about by the concept of Mutually Assured Destruction (MAD) that has prevented a nuclear war.
The authors caution that AI superintelligence—which they define as “AI vastly better than humans at nearly all cognitive tasks”—will pose grave risks to humanity in the coming years. They don’t mince words, writing, “‘Superintelligent’ AI surpassing humans in nearly every domain would amount to the most precarious technological development since the nuclear bomb.”
Well, here’s one scenario they describe regarding advanced AI-developed weapons:
Subnuclear superweapons—such as an AI-enabled cyberweapon that can suddenly and comprehensively destroy a state’s critical infrastructure, exotic [electromagnetic pulse] devices, and next-generation drones—could confer sweeping advantages without nullifying an adversary’s nuclear deterrent. Some superweapons might erode mutual assured destruction outright. A ‘transparent ocean’ would threaten submarine stealth, revealing the location of nuclear submarines. AIs might be able to pinpoint all hardened mobile nuclear launchers, further undermining the nuclear triad. AIs could undermine situational awareness and sow confusion by generating elaborate deceptions—a ‘fog of war machine’—that mask true intentions or capabilities. A defensive superweapon possibility is an anti-ballistic missile system that eliminates an adversary’s ability to strike back. Lastly, some superweapons remain beyond today’s foresight—’unknown unknowns’ that could undermine strategic stability.
Some suggest bolstering the US chip industry and working to impede other nation’s abilities to build a superintelligent AI, essentially.
In summary they opine that by detecting and deterring destabilizing AI projects through intelligence operations and targeted disruption, restricting access to AI chips and capabilities for malicious actors through strict controls, and guaranteeing a stable AI supply chain by investing in domestic chip manufacturing, states can safeguard their security while opening the door to unprecedented prosperity. Hmmmmm
The stakes couldn’t be higher but there is a precedent here, opponents of US AI Supremacy claim to site precedence: Nuclear annihilation has been avoided thus far despite significant tensions between nuclear states in part through the deterrence principle of Mutual Assured Destruction (MAD), where any nuclear use would provoke an in-kind response. In the AI era, a parallel form of deterrence could emerge—what might be termed ‘Mutual Assured AI Malfunction’ (MAIM)—where states’ AI projects are constrained by mutual threats of sabotage.
Comments(1)