Charles Lyons-Burt
Some of the biggest impediments to the adoption of artificial intelligence technologies across the Department of Defense are lack of trust in the technologies and resistance to them amongst the general public. According to William Streilein, chief technology officer at the Chief Digital and AI Office, these lingering doubts can be dissuaded through training and informational toolkits that help potential users of automated tech understand that they are a utility, rather than a threat.
“It’s a capability, it’s a technology. It’s not SkyNet,” Streilein told an audience of largely government contracting industry members at the Technology and Innovation Summit, hosted by ExecutiveBiz at 2941 Restaurant on October 6. (Streilien was referencing the AI antagonists of the Terminator film series.)
AI adoption is important to Streilein because its promotion is at the heart of the CDAO’s mission. The organization strives to “accelerate the DOD adoption of data, analytics and AI,” all three of which the CTO deems as “necessary to generate decision advantage from the boardroom to the battlefield.”
The CDAO is the result of the combination of four formerly individual organizations — Advana (“a business intelligence and business analytics platform,” per Streilien), the DOD’s Joint Artificial Intelligence Center, the Defense Digital Services (a crisis scenario actor) and the Chief Data Office. Thus, data is a key aspect of the CDAO’s purview, because, as Streilein explained, data centricity is intrinsically linked to responsible and effective AI implementation.
“The point of data centricity is not only to leverage data as an asset for your organization, but to empower everybody in your organization. And it means that you don’t require a decision to go all the way up the chain and come back down,” Streilein elucidated.
He went on to demonstrate how the empirical nature of data means that shaping one’s organization around it is a concrete and reliable methodology.
Also included in the roadblocks that prevent the acceptance or usage of AI technologies is inaccessible data. The CDAO CTO shared that this is due to a number of factors, such as policy barriers that outlaw certain use cases; metaphorical yellow tape intended to keep resources secure; or technical impediments like unreadable data formatting or antiquated numerology.
Acquisition and technical domain challenges comprise yet more ways that automated technology installations can be stymied. As does siloed or fragmented capability development, wherein innovation occurs but only disparately and disconnectedly.
“I can build a language translator probably overnight with the tooling that’s there. But is it integrated with other people and with what other people are doing? Can it be pulled together in that sort of similar integrated way so that we can leverage it there?” Streilein wondered illustratively.
Potential strategies for pushing back against these dissenting forces are certain enterprise capabilities, algorithmic warfare with a surge approach and conducting modern, agile acquisitions. Putting proper policy measures in place is also crucial, Streilein says, “to dictate and demand and set the rules so that we can leverage AI.”
In addition, the CDAO leader recommends responsible AI preparation toolkits that will educate discerning technologists on how to ethically build and maintain automated systems. In fact, Streilein quipped that he frequently considers that his employer could be called the Chief Digital and Responsible AI Office, “because we don’t do any other kind of AI, right?”
At the summit, Streilein additionally referenced another principle CDAO objective: supporting the Joint Forces and their shared combat capabilities. In this line of action, the agency will foster and execute AI tools that will impact the point at which “all the services come together to affect the warfighting capability.”
Not only are the CDAO’s efforts intended to assist and enable the Joint Forces, but to overall surpass the progress in AI and machine learning gained by U.S. adversaries.
