As the first-ever chief data officer at the U.S. Special Operations Command, or SOCOM, David Spirk Jr. is tasked with managing the high expectations of the Tampa, Florida-based organization. He is now celebrating his one-year anniversary on the job.
In a May speech, Gen. Richard Clarke Clarke, the new head of SOCOM, described how machine learning and AI can provide operators with tactical advantages by increasing situational awareness, reducing cognitive loads and improving the decision-making process.
“There are new, innovative ways to bring this reality to the forefront,” Spirk told C4ISRNET.
One of the most relevant SOCOM programs that could significantly benefit from machine learning and AI is the Hyper Enabled Operator concept. HEO replaces the five-year Tactical Assault Light Operator Suit effort.
“The HEO concept is one of the most important things SOCOM can be doing right now,” Spirk said. “The only way we have HEO in the future is if we can reduce the cognitive load of that operator and his ability to command his resources at his disposal through automation and applied AI.”
For example, machine learning and AI could help to process, exploit and disseminate data from intelligence, surveillance, and reconnaissance. Such an advantage could come in the form of new displays available to operators in the field. That technology could also improve automation at tactical operation centers, in turn providing special operations forces with mission-specific predictions, Spirk noted.
Before joining SOCOM, Spirk worked with the Department of Defense’s Project Maven program, which was started in April 2017 to design computer-vision algorithms for the processing, exploitation and dissemination of full-motion video.
The command is working to exploit machine learning and AI technologies across an increasingly complex operating environment, and Spirk emphasized SOCOM’s mission to adhere to special operations forces’ AI principles: grow the discipline, find a purpose-driven application and make it the norm.
“Quick wins are demanded,” Spirk said. SOCOM’s Command Data Office completed its first program review July 24 — an event Clarke attended.
Previous leaders also gave him “explicit guidance to go faster and deeper while responsibly challenging legacy policies or processes that get in the way of our progress,” Spirk said.
Such a strategy, he warned, is dependent upon cooperation across SOCOM; the Command Data Office in particular is tasked with identifying the most optimal means of creating an AI-ready workforce.
“Generating a new workforce to execute these concepts is important, providing a conduit to the force with individuals in our formation having begun learning about AI and data science to actualize it,” Spirk said.
He highlighted the critical nature of using special operations forces to identify “really creative ways to use technology which already exists today.”
“It’s about understanding which computer capabilities won’t take an algorithm and which workflows might benefit from AI,” he added.
As a result, Spirk is overseeing a series of projects aligned against Clarke’s applied AI focus areas:
- “Perception and action,” which will provide and receive support to and from Project Maven;
- “Planning and maneuver” opportunities that exploit commercial and academic wargaming techniques through the application of game theory;
- “Communications resilience and cyberwarfare” in order to protect the entire force;
- “Recruitment, training and talent management,” which is focused on AI to aid the assessment and selection of special operators;
- “Business processes,” which optimizes contract and budget management; and
- “Predictive maintenance and logistics,” which will exploit data science to build algorithms for predicting engine failures and maintaining high readiness levels starting with rotary-wing aircraft. This last effort will involve Carnegie Mellon University and the Joint Artificial Intelligence Center.
“We are talking about starting small with big intent. Take that bite, and once we show the art of the possible, stretch that win across the whole workflow,” Spirk said.
In September, SOCOM will hold its inaugural AI symposium to create a road map to build a sustainable AI portfolio.
“This is a huge opportunity to begin collecting operator responses while they’re under the headset, and to take the last 10 years of combat experience and make it structured, make it explicit to know exactly how operators are going to be acting in that environment,” Spirk said.
To this end, the Command Data Office will receive support from the forthcoming Data Engineering Laboratory co-located with the SOFWERX innovation in Ybor City, a neighborhood in Tampa; as well as staffers from the JAIC and the Defense Digital Service. The symposium will also consider how machine learning and AI can support SOCOM’s component commands, as well as theater special operations commands, Spirk said.
“We will prioritize SOF activities and resources, and challenge inhibiting practices to aid the development of a data-driven force of the future with purpose-driven applications,” he added.
Spirk reaffirmed Clarke’s priorities to compete and win for the nation, preserve and grow readiness, innovate for future threats, advance partnerships, and strengthen the force.