According to a new RAND Corporation paper, AI could potentially overturn the foundation of nuclear deterrence by 2040. The dangers of AI for nuclear security are based on AI’s potential to push humans toward incredibly dangerous risks.
Source: U.S. Army
The RAND paper says that AI could potentially eliminate the current code of mutually assured destruction that has been generally accepted by all world leaders since the end of the Cold War. New technology, like improved sensor technologies, could make it easier for military items like submarines and mobile missiles to be targeted and destroyed by other militaries and AI could push humans to use this new tech.
AI could tempt world leaders to engage in first-strike moves. By engaging in first-strike moves, countries could gain leverage over their enemies, even if they don’t want to engage in a further war or military attack later. First-strike movements undermine strategic abilities of countries that possess an area of land for military gains, even if the other military has no desire to use the land. But these movements could kill innocent civilians and anger opposing countries.
"The connection between nuclear war and artificial intelligence is not new, in fact, the two have an intertwined history," said Edward Geist, co-author on the paper and associate policy researcher at the RAND Corporation, a nonprofit, nonpartisan research organization. "Much of the early development of AI was done in support of military efforts or with military objectives in mind."
Geist noted that there is one example of this kind of work, the Survivable Adaptive Planning Experiment that was launched in the 1980s. This experiment aimed to use AI to translate reconnaissance data into nuclear targeting plans.
The researchers say that AI isn’t all bad, though. It could enhance strategic stability through improved intelligence collection and analysis, increase second-strike force vulnerability, improved analytics for monitoring and interpret adversary actions. All of this could reduce miscalculation in a military setting and stop an escalation of a dangerous military situation.
With more improvements, it is possible for AI systems to develop to be less error-prone than humans and could be more stable than humans in the long term. This means AI could eliminate humans altogether, which would also be eliminating the emotions from military decisions where people’s lives are at stake.
"Some experts fear that an increased reliance on artificial intelligence can lead to new types of catastrophic mistakes," said Andrew Lohn, co-author of the paper and associate engineer at RAND. "There may be pressure to use AI before it is technologically mature, or it may be susceptible to adversarial subversion. Therefore, maintaining strategic stability in coming decades may prove extremely difficult and all nuclear powers must participate in the cultivation of institutions to help limit nuclear risk."
The study is available on the RAND site.
