AI Shouldn't be Allowed to Push a Nuclear Button – Military Expert

There is a good reason for that, given that the two powers are increasingly competing in the AI field, Alexei Leonkov, a veteran Russian military analyst and editor of the magazine Arsenal of the Fatherland, told Sputnik.

Earlier this year, researchers from the Georgia Institute of Technology, Stanford University, Northeastern University, and the Hoover Wargaming and Crisis Simulation Initiative, tested several AI models as decision-makers in war simulations and concluded that AI had a tendency to start arms race and nuclear confrontations in different scenarios.

Leonkov notes that the US military started to use AI in the army from 2018. It was expected that the tool could be integrated into nuclear deterrence architecture, including intelligence, surveillance and reconnaissance systems and even nuclear command, control and communications.

However, authorities soon realized that AI is prone to errors and can be disrupted by cyber-attacks or confused by incoming contradictory data.

It also has limitations when it comes to decision-making in a chaotic battlefield environment, the expert points out. "In these conditions, artificial intelligence does not work at all," he says.

"Therefore, putting [AI] into service, given all the shortcomings that it has, is pure madness," the pundit says.