In the Middle East, the US uses AI for its bombings

On February 2, the US launched a series of strikes on weapons depots and militant command centers in Iraq and Syria. Iran-backed agents were blamed for the U.S. response to a drone strike in Jordan in January that killed three U.S. soldiers. In total, more than 80 bombings during this mission in the Middle East, and Bloomberg The Pentagon reveals that it relied on artificial intelligence (AI) to detect its targets, with a good dose of machine learning to teach it how to recognize a potential target. Information confirmed by Schuyler Moore, Chief of Technology, United States Central Command. “We use computer vision to identify potential threats. This method includes training algorithms that visually identify specific objects. »

Controversy at Google

These mechanisms were developed as part of Project Maven, which was launched in 2017 to improve automation within the Ministry of Defence. The Pentagon used them to locate rocket launchers in Yemen and fighter jets in the Red Sea, which it later destroyed in several airstrikes that same month.

Initially, the project relied on Google's artificial intelligence (AI) technology to analyze drone images and flag images for further human review. This caused turmoil among Google employees, with some resigning. As a result, Google did not renew the contract when the two-year contract expired.

Read more >> AI Solves Nuclear Fusion Puzzle

AI doesn't pull the trigger

This withdrawal did not stop the US from continuing this work, and it accelerated with Israel's attack on Gaza and its ramifications throughout the region. However, the US notes that AI is not used to trigger strikes and that players must verify every recommendation proposed by the algorithms. “There's no process that ever works, comes to a conclusion, and moves on to the next step,” says Schuyler Moore. “Every step involving AI is completed by a human. »