Error-Prone AI Nearly Sparked World War Three Last Month

In an article for The Guardian, scholars Timnit Gebru and Emily M. Bender argue that policymakers are excessively concentrating on hypothetical superintelligence while neglecting the more pressing issue of using unreliable AI in high-stakes situations. They cite an unconfirmed CNN report claiming that U.S. officials could have prevented a confrontation with a Chinese vessel following an AI-generated intelligence report that mistakenly stated the ship was carrying nuclear weapons components.

The article states, "While many of us were preoccupied with fictional superintelligent machines that could hypothetically cause human extinction, there was a more serious story unfolding." According to the report published by CNN on September 18, which has not been verified by other major sources, U.S. military personnel were prepared to intercept a Chinese vessel based on an intelligence report that identified the ship as carrying nuclear weapon components.

The military was ready to deploy aircraft and soldiers to board the ship. The article notes, "A military skirmish of this nature between China and the US could quickly escalate and spiral into world war three." It was only after further investigation that officials discovered the intelligence report was produced using a chatbot and contained false information about the vessel's contents.

The article concludes, "Unlike the terminator scenario, this incident did not prompt lawmakers to call for regulation to prevent potential human extinction. Media outlets failed to inform the public that we nearly triggered a war against a nuclear superpower due to our military's reliance on error-prone chatbots."