Doomsday Machine: Can Artificial Intelligence Start a Nuclear War?
Stanford University simulations have shown that current artificial intelligence models are prone to escalating conflicts to the point of nuclear weapons. The study raises serious questions about the risks of automating military decisions and the role of AI in future wars.
This is reported by Politico .
The results of war games conducted by researcher Jacqueline Schneider from Stanford indicate that artificial intelligence could become a dangerous factor in modern wars if it gains influence over military decision-making.
According to the scientist, during simulations, the latest AI models consistently chose aggressive escalation scenarios, including the use of nuclear weapons. Schneider compared the behavior of the algorithms to the approach of Cold War general Curtis LeMay, who was known for his willingness to use nuclear force on minimal pretext.
“Artificial intelligence models understand perfectly well how to escalate a conflict, but are actually unable to offer options for its de-escalation,” the researcher explained.
In her opinion, this is due to the fact that most of the military literature used to train AI describes escalation scenarios, not those that avoided war.
The Pentagon assures that AI will not have the right to make decisions about launching nuclear missiles, and emphasizes the preservation of “human control.” At the same time, modern warfare is increasingly dependent on automated systems. Already today, projects like Project Maven rely entirely on machine-generated intelligence data, and in the future, algorithms will even be able to advise on countermeasures.
In addition, there are already examples of automation in the field of nuclear weapons in the world. Russia has the Perimeter system, capable of delivering a strike without human intervention, and China is investing huge resources in the development of military artificial intelligence.
Journalists also recall the case of 1979, when US President Jimmy Carter’s advisor Zbigniew Brzezinski received a message about the alleged launch of 200 Soviet missiles. Only a moment before the decision to retaliate was made, it turned out that this was a system error. The question is whether artificial intelligence, which works “reflexively”, would have been able to wait for more detailed information, or would have pressed the “red button” automatically.
Thus, the discussion about the role of AI in the military sphere is becoming increasingly relevant, because not only the outcome of the battle, but also the fate of all humanity may be at stake.
It was previously reported that the Thwaites Glacier in Antarctica, nicknamed the “Doomsday Glacier,” is losing stability and could trigger a rapid rise in sea levels by several meters.
Recall that Hollywood actor and musician Will Smith was involved in a scandal. In particular, the star of “Men in Black” was suspected of using artificial intelligence.
Also follow “Pryamim” on Facebook , Twitter , Telegram , and Instagram.