A recent study by the RAND Corporation suggests that artificial intelligence could increase the risk of nuclear war by the year 2040.
That would not happen because of AI taking control and launching missiles as it pleases, but as a result of the influence the information it provides could have on decisions involving military engagement.
“During the Cold War, the condition of mutual assured destruction maintained an uneasy peace between the superpowers by ensuring that any attack would be met by a devastating retaliation,” a release from RAND states. “Mutual assured destruction thereby encouraged strategic stability by reducing the incentives for either country to take actions that might escalate into a nuclear war.”
“In coming decades, artificial intelligence has the potential to erode the condition of mutual assured destruction and undermine strategic stability,” the release further notes.
One example is through its enhanced ability to find, and even create, vulnerabilities in a nation’s strike response plans and resources.
Knowing that such possibilities exist could create defensive concerns and the conclusion that the best form of self-preservation is to attack first.
There is also the risk that AI will be given more responsibility than it is ready to take on.
“There may be pressure to use AI before it is technologically mature, or it may be susceptible to adversarial subversion,” Andrew Lohn, one of the paper’s authors, commented. “Therefore, maintaining strategic stability in coming decades may prove extremely difficult and all nuclearpowers must participate in the cultivation of institutions to help limit nuclear risk.”