We saw a demo of the new AI system powering Anduril’s vision for war

Currently, humans also have to translate between systems made by different manufacturers. A soldier might have to manually rotate a camera to look around the base to see if there is a drone threat, and then manually send information about the drone to another soldier operating a weapon to shoot it down. These commands can be shared through a low-tech messaging application – comparable to AOL Instant Messenger. This takes time. This is the problem the Pentagon is trying to solve with its solution All-domain joint command and control Plans and other measures.

“We’ve known for a long time that our military systems don’t interoperate,” said Chris Burrows, former staff director of the Senate Armed Services Committee, chief adviser to Sen. John McCain, and now Anduril’s chief strategist. Much of his job has been to convince Congress and the Pentagon that software issues deserve as much a place in the defense budget as jets and aircraft carriers. (Anduril spent nearly $1.6 million on lobbying last year, according to Open Secrets, and has close ties to the incoming Trump administration: Anduril founder Palmer Luckey has been a long-time Trump donor and supporter, with J.D. Vance leading the charge After investing in Anduril in 2017, he worked at the venture capital firm Revolution.

Defense hardware also faces connectivity issues. Tom Keane, Anduril’s senior vice president of connected warfare, told me a simple example from the civilian world. If you receive a text message when your phone is turned off, you will see the message when you turn your phone back on. It was preserved. “But this functionality we didn’t even think about,” Keane said, “doesn’t really exist” in the design of many defense hardware systems. In challenging military networks, data and communications can easily be lost. Anduril said its system stores data locally.

AI data treasure trove

The push to build more artificial intelligence-connected hardware systems in the military could lead to one of the Pentagon’s largest data collection projects ever, and companies like Anduril and Palantir also have big plans.

“Exabytes of defense data essential for AI training and inference are disappearing,” Anduril explain On December 6, the company announced that it would work with Palantir to compile data collected by Lattice (including highly sensitive confidential information) to train artificial intelligence models. Training on the broader data collected by all these sensors will also greatly facilitate Anduril’s current model-building work with OpenAI. declare December 4th. supply Its artificial intelligence tools can help the Pentagon reimagine how it classifies and manages classified information. When Anduril founder Palmer Luckey told me in an October interview that “there’s not a lot of information on classified topics and weapons systems understanding” to train artificial intelligence models, he may have been foreshadowing what Anduril is now building.

Even if the military already collects some data, artificial intelligence will suddenly make it more useful. “What’s new is that the Department of Defense now has the ability to use this data in new ways,” Emilia Probasco, a senior fellow at Georgetown University’s Center for Security and Emerging Technologies, wrote in an email. “More More data and the ability to process data can support greater accuracy and precision and faster information processing.”

The sum of these developments may be that AI models can be more directly incorporated into military decision-making. The idea has drawn scrutiny since last year it was discovered that Israel had been using advanced artificial intelligence models to process intelligence data and produce target lists. Human Rights Watch wrote in an article Report These tools “rely on faulty data and imprecise approximations.”

“I think we are already on the path to integrating artificial intelligence, including generative artificial intelligence, into decision-making,” Probasco said. Recent analysis One such case. She studied a system built within the military in 2023 called the Maven Intelligence System, which allows users to “access sensor data from different sources. [and] Apply computer vision algorithms to help soldiers identify and select military targets.