Free beta access. 50 free AI tokens each week. Optional top-ups available. See what is included
Opening Word document…

AI-controlled nuclear weapons auto-launch at the slightest provocation.

John Allen 21, president of the Brookings Institution, distinguished fellow in the Foreign Policy Program at Brookings; Darrell West, vice president and director of Governance Studies and holds the Douglas Dillon Chair in Governmental Studies, 3/24/2021, "It is time to negotiate global treaties on artificial intelligence," https://www.brookings.edu/blog/techtank/2021/03/24/it-is-time-to-negotiate-global-treaties-on-artificial-intelligence/, sg

With the rise of artificial intelligence, supercomputing, and data analytics, the world today is at a crucial turning point in the national security and the conduct of war. Sometimes known as the AI triad, these characteristics and other weapons systems, such as hypersonics, are accelerating both the speed with which warfare is waged, and the speed with which warfare can escalate. Called “hyperwar” by Amir Husain and one of us (John R. Allen), this new form of warfare will feature levels of autonomy, including the potential for lethal autonomous weapons without humans being in the loop on decision-making.

How artificial intelligence is transforming the world

Assessing ethical AI principles in defense

It will affect both the nature and character of war and usher in new risks for humanity. As noted in our recent AI book “Turning Point,” this emerging reality could feature swarms of drones that may overwhelm aircraft carriers, cyberattacks on critical infrastructure, AI-guided nuclear weapons, and hypersonic missiles that automatically launch when satellite sensors detect ominous actions by adversaries. It may seem to be a dystopian future, but some of these capabilities are with us now. And to be clear, both of us, and more broadly the world’s liberal democracies, are struggling with the moral and ethical implications of fully autonomous, lethal weapon systems.

In this high-risk era, it is now time to negotiate global agreements governing the conduct of war during the early adoption and adaptation of AI and emerging technologies to the waging of war and to specific systems and weapons. It will be much easier to do this before AI capabilities are fully fielded and embedded in military planning. Similar to earlier treaties on nuclear, biological, and chemical weapons in the post-war period, these agreements should focus on several key principles:

Incorporate ethical principles such as human rights, accountability, and civilian protection in AI-based military decisions. Policymakers should ensure there is no race to the bottom that allows technology to dictate military applications as opposed to basic human values.

Keep humans in the loop with autonomous weapons systems. It is vital that people make the ultimate decisions on missile launches, drone attacks, and large-scale military actions. Good judgment and wisdom cannot be automated and AI cannot incorporate necessary ethical principles into its assessments.

Adopt a norm of not having AI algorithms within nuclear operational command and control systems. The risk of global destruction is high with AI-based launch on warning systems. Since we do not know, and may never know, exactly how AI learns from training data, it is important not to deploy systems that could create an existential threat to humanity.

Sign in to recut, update, or save this card.