The recent buzz over artificial intelligence has reignited fears that the technology could pose an existential threat to humanity. The latest warnings evoke dystopian scenarios of AI surpassing human control, reminiscent of Hollywood blockbusters like The Terminator. But as policymakers grapple with these concerns, they can look to historical precedents—climate change, pandemics, and thermonuclear weapons—for guidance on how to manage such a formidable force.
Lessons from Past Existential Threats
Climate change, for instance, has triggered ice ages and could bring catastrophic warming. While severe, efforts to curb greenhouse gases and pollutants have helped mitigate some of the worst outcomes. Similarly, the threat of massive pandemics cannot be dismissed; the possibility of AI-engineered diseases, such as the fictional White Plague from Frank Herbert's 1982 novel, looms large.
But perhaps the most instructive parallel is the development of thermonuclear weapons. The creation of these weapons, which are exponentially more powerful than those dropped on Hiroshima and Nagasaki, led to existential dread. J. Robert Oppenheimer, the father of the atomic bomb, famously quoted the Bhagavad Gita, expressing his profound pessimism about humanity's future. Yet, thermonuclear war has been avoided, partly due to a bifurcated control system that separated military and civilian oversight. The Atomic Energy Commission, established by President Truman in 1947, and later the Nuclear Regulatory Commission, along with the Departments of Defense and Energy, have managed the nuclear complex.
AI's Unique Dangers
Unlike nuclear weapons, which require vast infrastructure, AI is more accessible and could be exploited by individuals. It could facilitate the creation of man-made pandemics or disrupt critical infrastructure like power and water, threatening food supplies. While AI in its current form may not build a nuclear weapon, it could be used to tamper with existing ones, and hacks against AI systems are already frequent.
Proposed Federal Framework
To address these risks, experts like Harlan Ullman, a defense strategist and principal author of the doctrine of shock and awe, advocate for a federal regulatory agency for AI. This agency would establish guidelines and oversee the industry, similar to the Nuclear Regulatory Commission. The degree of regulation would be a political decision, but the model could be replicated internationally under the United Nations, akin to the International Atomic Energy Agency.
In the U.S., the Departments of Defense, Energy, and Commerce, along with the CIA, would assess national security threats from AI and determine appropriate responses. Negotiations on AI, similar to arms control agreements, could follow, though they would be more complex given AI's widespread availability.
Ullman, whose upcoming book Who Thinks Best Wins co-authored with former UK Chief of Defense Lord David Richards, argues that with the right foundations and leadership, we can prevent an AI catastrophe, just as we have avoided thermonuclear war. The question is whether policymakers will act with the same urgency they applied to the nuclear age.
As the debate over AI regulation gains momentum, lawmakers are also considering related issues, such as AI's impact on middle-class jobs and the threat to artists' livelihoods. These concerns highlight the need for a comprehensive approach to AI governance that addresses both existential risks and economic disruptions.
