The Perils of AI for Nuclear Deterrence

Abstract

The use of artificial intelligence (AI) in nuclear affairs could be potentially dangerous. Quantum computing and AI based detection systems, speedy delivery platforms, revolution in remote sensing and precision targeting will challenge hardening, concealment, and survivability of the command and control, delivery platforms, weapons, decision makers and strategic assets.1 Therefore, the future nuclear forces and arsenals are likely to be more ready to use, decision making time urgent and nuclear doctrines are likely to be conceptually aggressive. However, AI inferior states may increase the number of their nuclear warheads to ensure a retaliatory strike, thus raising proliferation concerns. Moreover, these states will also be compelled to use cyber offences and time urgent postures as countermeasures. On the other hand, AI superior states may resort to counterforce first strike to ensure superiority in a nuclear war. This study will particularly explore how AI would increase the risk of a nuclear war, and impact the nuclear deterrence concepts.