U.S. Army Gen. Richard D. Clarke, U.S. Special Operations Command commanding general, surveys an urban combat training complex at Melrose Air Force Range, N.M., September 9, 2020. U.S. Air Force photo by Senior Airman Maxwell Daigle
SOCOM Members Got an All-Star Crash Course in AI
Over a unique six-week course, participants met virtually with tech leaders to talk about AI's future.
Troops and civilians with U.S. Special Operations Command participated in a unique course this summer with MIT academics and other technology leaders to discuss AI and how it might shape the future of combat and other areas of human activity.
The six-week course allowed more than 300 SOCOM representatives, including members of senior leadership as well as other, lower-ranked officers and civilians, to understand broad future trends in AI development, commercial deployment, and more.
“What we did for them is bring together at MIT some of the best known thought leaders in the world of AI and have them provide their unique perspective,” Sertac Karaman, a professor in the Department of Aeronautics and Astronautics at MIT and one of the course’s organizers, told Defense One. Those thought leaders included longtime Google CEO Eric Schmidt and former Defense Secretary Ash Carter.
The discussion included how operators might use tools like computer vision in the near future, “not just to detect vehicles here or there, but try to make sense of the imagery in a conceptual way,” Karaman said.
They also discussed other emerging AI capabilities, like the near future of natural language processing, which is a key focus area for SOCOM.
Other portions of the course were devoted to AI safety and how to pursue new capabilities while adhering to the Pentagon’s AI ethical principles. On Wednesday, SOCOM head Gen. Richard Clarke told lawmakers that one of the key features of the course from a SOCOM perspective was to help senior and mid-tier SOCOM personnel “learn what the AI principles were so that they could ask the right questions.”
The participants “were very interested in the robustness and ethical aspects… they were very careful with: How do they take AI and fit it into the ethical framework that they already have at DOD?” Karaman said.
But the participants also had a lot of questions about how high-tech adversaries might deploy AI in new and surprising ways, particularly on future battlefields against adversaries not bound by any ethical guidelines. Questions like, “What is the best way to counter that? How do we understand what kind of capabilities [adversaries] might deploy? How do we stay ahead so that we don’t get… blindsided by our adversaries?” Karaman said.
It’s the sort of course that could expand to other areas of the Defense Department, where Karaman hopes it will go on to influence aspects of acquisition, one of the key points he made during the course.
“I did tell them, for example, China is able to acquire any technology that’s built by any company inside China, but that’s not the case for the United States,” he said.
“We [meaning the United States] are still quite far ahead in terms of deploying some real AI capability that would really tilt the scale. But there’s a whole issue of integrating new technologies into DOD.”
It’s an area where the Defense Department must urgently catch up to the commercial world, he said.