SOCOM Says Learning to Leverage Data is Key for AI

U.S. Special Operations Command (SOCOM) and the U.S. Army see data standards and governance as the first step toward AI excellence and successful implementation of the Defense Department’s Joint All Domain Command-and-Control (JADC2) initiative.

Douglas Matty, Director of AI Capabilities at the Army, said AI will be a warfighter enabler, which requires a holistic approach to data and AI.

“When it comes to AI we think about the AI stack,” he said during the 2022 DOD Digital & AI Symposium Wednesday. “It’s more than just an algorithm. How are the sensors, data management, algorithms, human interaction with the results of that capability? Our biggest takeaway in 2021 [from Project Convergence] was it really has to be joint as well as allies and partners coming together to achieve convergence of multidomain operations.”

Maj. Ryan Harth, AI Division Deputy Director in the Force Modernization Center at U.S. Army Special Operations Command (USASOC), told GovCIO Media & Research in an interview that his primary focus is leveraging USASOC data for prescriptive and descriptive analysis, which requires data standards and governance.

“The real focus is on enabling elements within the enterprise to create and deploy responsible AI,” Harth said in the interview. “AI is at the center of the JADC2 framework. The [USASOC] AI Division has a strong focus on architecting solutions that facilitate this versus prescribing one piece of software, this focuses on enabling organizationss within the USASOC to leverage the data.”

Effective data management is at the heart of a recent successful AI use case where a USASOC team used natural language processing (NLP) to support vehicle maintenance in theater.

“If the machines and equipment don’t work it’s very hard for operators to accomplish their mission,” Harth said. “That program has been able to identify ways to improve the efficiency of various systems within the maintenance field and save money and save time, which directly translates to better support to the warfighter and individuals performing the maintenance on the equipment as well. If you can tell a maintainer to put their equipment in a different location to reduce the time to reach that piece of equipment, it decreases downtime and overall cost.”

Leveraging data efficiently and effectively is the heart of AI, Harth added, and successful implementation relies on a hybrid approach to the cloud.

“At the center of AI, ML, and predicative analytics is data,” he said. “You can only do any of those with enough of the right and or relevant data. For organizations that are geographically dispersed, such as USASOC, it’s almost mandatory to leverage a hybrid solution to architect the pipelines necessary to move from a producer to a production environment.”

SOCOM CDO Thomas Kenney said DOD’s data platform, ADVANA (which is now run by the new Chief Digital and AI Office) is helping SOCOM collaborate more effectively with the other service branches, which will help identify more AI use cases.

“And part of the data-centric mindset and evolving into the AI-enabled command is educating the force we’ve committed to programs on MIT and Carnegie Mellon to educate senior leaders to get their mindset to change into how this can effectively be used in human machine teaming,” Kenney said at the AI symposium Wednesday.

Developing a data-centric mindset and shifting to an AI-enabled command requires a culture shift, which DOD’s new CDAO, Craig Martell, discussed in a fireside chat with DOD CIO John Sherman.

Harth believes DOD will eventually evolve into an AI-friendly culture and should focus on educating the workforce to achieve a culture that fosters healthy data practices and AI innovation. He’s not so much concerned about attracting AI talent as he is about upskilling and shepherding current employees into a data-enabled future.

“Becoming an organization that is capable of leveraging the data that is produced, because it’s a very large amount, is key,” he said. “Data is produced, we want to analyze it, we want to enable folks to make decisions [so] they can leverage that data and improve processes. There’s tremendous data in the ecosphere right now. The talent piece is not necessarily the problem, there’s tremendous talent across the organization. As this generation of servicemembers and gov civilians have the opportunity to move into leadership roles, I think that in itself will address the cultural nuances at play for organizations working towards becoming data-driven.”