At a defense forum in Singapore, a discussion of strategic stability shifted in large part from nuclear weapons to the risks posed by artificial intelligence. Panelists warned that AI could shorten decision-making times so dramatically that people may act hastily and become more likely to make mistakes.
At issue is the compression of the so-called OODA loop—observe, orient, decide and act. According to Lieutenant General Nauman Zakria, commander of Pakistan’s 1 Corps and Army Rocket Force Command, this creates a situation in which “a human cannot assess the situation quickly enough.”
“People will act irrationally, and those actions will be extreme,” he said on Saturday during a discussion at the Shangri-La Dialogue.
Similar concerns were voiced by the only panelist not working directly in defense or security—Mirjana Spoljaric, president of the International Committee of the Red Cross. She noted that technology can improve people’s lives, but at the same time sharply increases the risks of war.
“We don’t know where the trigger is pulled,” Spoljaric said. “It could be happening thousands of kilometers away. So while AI may have the potential to help protect civilians, what we see now is only its negative side.”
Nuclear weapons—the classic example of strategic deterrence—also remained an important part of the discussion. Major General Meng Xiangqing of China’s People’s Liberation Army reaffirmed Beijing’s commitment to a no-first-use nuclear policy and said the world would be safer if more countries adopted such an approach.
“If we can do this, we can reduce risk and further strengthen strategic stability,” he said.
But the conversation returned again and again to artificial intelligence.
General Onno Eichelsheim, chief of defense of the Netherlands, noted that such technologies are already being used on the battlefield: Ukraine uses them to forecast Russian attacks. The United States has also said it used AI in planning strikes on Iran.
“AI is a huge risk in terms of escalation. I think that is obvious,” Eichelsheim said. “But I am not naive. It will be used in this field. It already is being used.”