U.S. Media: Concerns Arise Over U.S. Military's Overreliance on Artificial Intelligence on the Battlefield

Reference News Network reported on May 4 According to an Associated Press report from Washington on May 1, the Pentagon stated on the 1st that it has reached agreements with seven technology companies to apply these companies' artificial intelligence (AI) technologies to its classified computer networks, enabling the military to leverage AI-driven capabilities to assist in combat operations.

The U.S. Department of Defense stated that Google, Microsoft, Amazon Web Services, NVIDIA, the OpenAI Research Center, Anthropic, and SpaceX will provide resources to assist in "optimizing decision-making for combat personnel in complex operational environments."

Notably, Anthropic is not on this list. Previously, the company had public disputes and legal issues with the Trump administration over ethical and security concerns regarding the application of AI in warfare.

In recent years, the U.S. Department of Defense has been accelerating the application of artificial intelligence.

A report released by the Brennan Center for Justice in March this year stated that this technology could help the military reduce the time required to identify and strike targets on the battlefield, while also assisting in coordinating weapon maintenance and supply lines.

However, the application of artificial intelligence has raised concerns, including potential violations of the privacy of U.S. citizens or allowing machines to select targets on the battlefield. One of the companies contracted with the Pentagon stated that its agreement requires human oversight in specific situations.

During Israel's war against armed groups in Gaza and Lebanon, concerns began to arise regarding the military application of artificial intelligence, as major U.S. tech companies secretly provided technical support to Israel for target acquisition.

As civilian death tolls sharply increased, there was growing concern about these tools causing the deaths of innocent people.

Helen Toner, acting director of the Center for Security and Emerging Technology at Georgetown University, stated that as the Pentagon signs the latest agreements, there are concerns that the military may become overly reliant on this technology on the battlefield.

Toner, who previously served on the board of the OpenAI Research Center, said, "Modern warfare largely relies on personnel sitting in command centers monitoring screens, making various complex decisions in response to chaotic and rapidly changing situations. AI systems are helpful in summarizing information, viewing surveillance footage, and attempting to identify potential targets." However, she noted that the appropriate level of human involvement, associated risks, and personnel training are still under discussion.

Toner asked, "How can we quickly roll out these tools to make them effective and create a strategic advantage? At the same time, we must recognize the need to train operators to ensure they know how to use them and do not over-rely on these tools."

The Pentagon stated on the 1st that military personnel have already begun using various AI capabilities through official platforms.

The Pentagon said, "Combat personnel, civilians, and contractors are applying these capabilities in practice, reducing the time required to complete multiple tasks from months to days." The Pentagon also stated that the military's continuously improving AI capabilities will "provide combat personnel with the necessary tools to act confidently and defend the nation against any threats."

Toner indicated that in many cases, the military's use of AI is similar to that of the general public, using it to handle mechanical tasks that would take humans hours or even days to complete.

She stated that AI can more accurately predict when helicopters need maintenance and can plan how to efficiently transport large-scale troops and equipment. It can also assist in determining whether vehicles in drone surveillance footage are civilian or military. However, people should not over-rely on AI. "There is a phenomenon known as automation bias, where people's imagined effectiveness of machine use often exceeds actual performance." (Translated by Qing Songzhu)