As the future of warfare shifts toward artificial intelligence, Ukraine has a valuable resource: millions of hours of drone footage that can be used to train AI models to take decisions on the battlefield.

AI was deployed by both sides on the battlefield during Russia’s invasion of Ukraine to identify targets, scanning images much faster than a human.

Oleksandr Dmitriev, founder of OCHI, a Ukrainian nonprofit digital system that centralizes and analyzes video feeds from more than 15,000 drone teams working on the front lines, told Reuters his system had collected 2 million hours , or 228 years, of battlefield videos from drones since then. 2022.

This will provide vital data that AI can learn from.

“It’s food for the AI: if you want to teach an AI, you give it 2 million hours (of video), it will become something supernatural.”

According to Dmitriev, the images can be used to train AI models in combat tactics, target detection and assessing the effectiveness of weapon systems.

“It’s basically an experiment that can be turned into math,” he said, adding that an AI program can study the trajectories and angles at which weapons are most effective. The system was initially designed in 2022 to give military commanders an overview of their areas of the battlefield by showing them side-by-side drone footage of all nearby crews on a single screen.

After the system was deployed, the team running it realized that video sent back from drones could be useful for recording war. So they started storing them.

On average, Dmitriev said five to six terabytes of new data are added every day because of the fighting.

Image quality

Dmitriev said he was talking with representatives of some of Ukraine’s foreign allies who had expressed interest in its OCHI system, but declined to provide details.

Samuel Bendett, deputy senior fellow at the US-based Center for a New American Security, said such a vast pool of data would be extremely valuable in teaching AI systems to identify exactly what they see and what measures they must take.

“Humans can do this intuitively, but machines cannot, and they need to be trained to know what is or is not a route, a natural obstacle or an ambush,” he said.

Kateryna Bondar, a researcher at the Wadhwani AI Center at the Center for Strategic and International Studies, said the size of the dataset and the quality of the image were important as the AI ​​models learned to recognize targets based on shapes and colors.

Bondar said the data set was valuable in the context of training to fight Russia. However, she said U.S. officials and drone makers preferred a data set to train AI systems to operate in the Pacific against a potential Chinese adversary.

“(They want) systems that are ready and able to fight China because that’s the main priority for the United States right now, rather than getting a lot of images of Ukrainian fields and forests.”

Ukraine also has another system, called Avengers, developed by its Defense Ministry, which centralizes and collects drone and CCTV videos.

The ministry refused to provide information on this system. However, it has already been reported that Avengers spots 12,000 pieces of Russian equipment per week using AI identification tools.

Thousands of drones already use AI systems to hit targets without human piloting, and Ukraine is using AI technologies to help clear its territory.

Ukrainian companies are developing drone swarms, in which a computer system will be able to execute commands for an interconnected cloud of dozens of drones.

Russia has also touted its use of AI on the battlefield, including for target recognition for Lancet strike drones, which have proven deadly against Ukrainian armored vehicles.