The Department of Defense is majorly scaling up artificial intelligence in the military in the hopes of faster decision-making in warfare.
The Department of Defense’s Defense Innovation Unit awarded artificial intelligence giant Scale AI a prototype contract for “Thunderforge,” the department’s flagship program to integrate AI into military planning and operations.
“Thunderforge marks a decisive shift toward AI-powered, data-driven warfare, ensuring that U.S. forces can anticipate and respond to threats with speed and precision,” the Department said in a press release on Wednesday.
Although the financials of the contract were not disclosed, CNBC (CMCSA+1.81%) reported that it was a multi-million dollar deal.
Besides Scale AI, the Thunderforge system will also include defense company Anduril’s Lattice open software platform, and LLM technology by Microsoft (MSFT-1.25%).
The Department has been working to get AI capabilities into defense operations since 2021, including through the use of autonomous weapons and AI-powered computer vision to identify airstrike targets.
But Thunderforge is its first significant step into giving AI a more prominent role in operational decision-making across the military by integrating large language models, dubbed AI agents, in its workflows. AI will be used in military campaign development and resource allocation, wargaming simulations, planning scenarios and proposed courses of action, and strategic assessments.
U.S. Central Command’s chief tech officer Schuyler Moore told Bloomberg in 2024 that Centcom had experimented with AI recommendation engines in late 2023 and found that it “frequently fell short” of humans when proposing orders of attack.
The Department will first deploy the system within the U.S. Indo-Pacific Command (INDOPACOM), out of Hawaii, and the U.S. European Command (EUCOM), based in Germany, before scaling it across the rest of the eleven combatant commands.
Humans will oversee the AI agents, Scale AI shared in a press release on Wednesday, but that has not quelled worries over the technology being deployed in certain fields.
“We all probably suffer from automation bias, which is this idea that we are tempted to and often will accept the recommendation, for example, that a large language model spits out or prediction that one of these systems is making, because we feel as though the system must have more information than we do, and must be processing it and sequencing it and ordering it better than we could,” legal scholar and former associate White House counsel Ashley Deeks told Quartz.
What exacerbates the problem even more is that AI systems are like “black boxes,” according to Deeks, in that it is tough for users to understand how or why it reaches certain conclusions.
“I hope that the Pentagon itself is thinking about how to train people to resist excessive automation bias when their gut and their experience has told them to do ‘x’ and the system is telling them to do ‘y’,” Deeks said.
Scale AI CEO and founder Alexandr Wang has been praising the merits of AI-assisted warfare for some time now. Wang took out a full page ad in the Washington Post asking President Donald Trump to invest more to “win the AI war” in January, and then defended his opinion later during a February summit in Qatar, where he said he is concerned that China will use AI to “leapfrog” the military capacity of “Western powers.” China has reportedly made AI military power a strategic priority, although some experts believe Beijing still faces significant obstacles in taking full advantage of the technology.
A number of militaries around the world have used AI to assist their military operations and identify targets, most notably Israel in its war in Gaza, and by both sides of the Russia-Ukraine war.
The post The Pentagon is upping its bet on AI. Here’s what it means for the military appeared first on Quartz.
