Nuclear deterrence experts are concerned about the introduction of artificial intelligence into military decision-making because testing shows that it has a tendency to escalate conflicts to the point of global nuclear war.
Nuclear deterrence experts are concerned about the introduction of artificial intelligence into military decision-making because testing shows that it has a tendency to escalate conflicts to the point of global nuclear war.
As Futurism reports, experts worry that AI will either gain the power to launch nuclear weapons on its own, or that humans with such powers will become so dependent on its instructions that they will do so on its recommendation. They point out that in war games, AI has a tendency to escalate conflicts to the level of nuclear war that humans could settle.
«It’s almost as if AI understands escalation but doesn’t understand de-escalation. We don’t really know why that happens,» says Stanford’s Jacklyn Schneider, director of the university’s Hoover Wargaming and Crisis Simulation Initiative, which has tested AI systems’ responses to war games.
The Pentagon now insists that humans will always be involved in the process. «The administration supports the need to maintain human control over nuclear weapons,» a senior military official said.
However, experts fear that this could change when Russia or China integrate AI into their military control structures, or the US military decides to enter a nuclear war on the advice of imperfect AI systems.
«I’ve heard combat unit commanders say, ‘Hey, I need someone who can take all the results of military exercises and, when I’m in a [crisis] situation, give me a solution based on the AI interpretation,’» Schneider noted.
Experts acknowledge that their concerns are fueled by associations with Dr. Strangelove’s doomsday machine, the military response plan from War Games, and Skynet from The Terminator. However, they point out that «the prophetic images of these science fiction films are quickly becoming reality.»
Earlier, top managers of leading technology companies, including Sam Altman from OpenAI, and scientists working on artificial intelligence, signed a joint statement on the dangers of technology, equating it to nuclear war or a pandemic.
According to a former Google and Microsoft engineer, the probability of humanity becoming extinct due to artificial intelligence is «at least 95%» if it continues on its current path.
