Pentagon Wants to Use Social Media On the Battlefield
Artificial intelligence will weave open-source and satellite data into useful intelligence in real time, the Pentagon says.
It still takes the US military too long to turn social media and other open-source information into something that operators in the field can use. Artificial intelligence is going to change that, and give U.S. troops a distinct battlefield edge, says US Deputy Defense Secretary Bob Work.(pictured)
Take the 2014 downing of Malaysian flight MH17 over Ukraine. (A conventional investigation by a European Union Joint Investigation Team took more than a year to affix blame to pro-Russian separatists operating a Russian-made BUK surface-to-air missile.)
To test the current state of machine learning applied to open source intelligence, the Pentagon hired a data integration and geospatial intelligence company called Orbital Insight, a big-data analytics company with a focus on satellite imagery and geospatial data. Most of their business is commercial, for example, they, analyze pictures of parking lots from space to predict holiday sales trends.
The company quickly scanned all available open-source media and assembled a picture of evidence, and did it instantly. He used slides to tell the story to an audience at the Center for Strategic and International Studies: “On the lower left is a Twitter shot of MH17 taking off…The next one comes from ParisMatch.com. It is the picture of the Russian SA-11 launcher with a serial number on it, date and time stamped near the village where the shoot-down occurred; then on Bellingcat.com, the exact same SA-11, at the exact same location. Then there’s a Twitter shot of a contrail of a missile rising at the time of the shoot-down.
Then a rebel leader takes credit for the shoot down on VK.com. That was immediately taken down, by the way. Finally, on YouTube, there’s a picture of the exact same SA-11 with a missile rail that is now mysteriously empty going back into Russia. Learning machines did this without any human interaction.”
That sort of rapid insight will be key to winning future conflicts where not all the players wear uniforms, like the masked “little green men” who invaded the Crimean peninsula in 2014. The United States and other international observers believed that Russian military officers were working with the separatists, a charge Russia at first denied. The ambiguity surrounding the identity of the invaders inhibited a coordinated international response.
“This type of stuff will allow us new indications and warning in gray-zone operations, the little green men,” said Work. “Learning machines can say, ‘There is an influence operation going on. We don’t know who is doing it but here are the key themes of the influence operation.’ There will be new means of going after terrorists. There will be new means of operating against regional powers. There will be new ways of operating against great state powers. This is totally transferable across the range of operations.”
It’s an example of why next-generation machine learning and artificial intelligence are key to the Pentagon’s push for technologies that will secure military dominance over competing nations, sometimes referred to as the Third Offset.
The military, of course, already uses social media for operations. In 2015, Air Combat Command’s Gen. Hawk Carlisle described how it helped the Air Force pull off a precision strike against ISIS. The airmen were “combing through social media and they see some moron standing at this command,” Carlisle said. “So they do some work, long story short, about 22 hours later through that very building, three JDAMs take that entire building out.”
The goal now is to shorten the time from data collection to actionable intelligence from a day to immediately.
Some observers pointed out that each of those data sources, particularly Bellingcat, the international open source news collective, provided a good indication of an influence operation afoot, and were doing so in real time.