The phantom menace: artificial intelligence could start a nuclear war by 2040
The researchers calculated that the AI in the future could trigger one of the nuclear powers to launch a preemptive strike against another nation.
American non-profit organization the RAND Corporation published a study according to which artificial intelligence can potentially lead the world to nuclear war by 2040.
AI can undermine geopolitical stability and to disrupt the status of nuclear weapons as a deterrent.RAND Corporation
Analysts say that although for decades, a fragile peace was maintained with the idea that any nuclear attack could trigger mutually assured destruction, the potential of AI and machine learning to make decisions about military action may mean that the lack of critical thinking in AI can give the wrong assessment of the situation. “Some experts fear that increased dependence on artificial intelligence may lead to new types of catastrophic errors,” said Andrew Lons, co-author of the RAND.
NewsArtificial intelligence helped “blow up” the Washington
The study’s authors believe that future improvements in technology may lead to disruption of the balance of deterrent forces, such as submarines and mobile missiles. AI, used as a counselor, can persuade a person to decide on a preemptive strike against another nation, even if the opponent does not intend to attack, but the AI will see the opportunity to gain an advantage in the attack.
AI may be “susceptible to competition”. Therefore, the said LON, the preservation of stability in the coming decades can be extremely difficult, and all nuclear powers should participate in the creation of institutions that would help reduce nuclear risks.