Artificial Intelligence for Federal Contractors

What is the Artificial Intelligence Hierarchy of Needs?

During the summer of 2022, Dr. Martell and his team at CDAO performed an audit of existing and emerging technology across all levels of the DoD, partners, and industry. Because the DoD is big, diverse, and largely decentralized, the CDAO connected with subject matter experts and stakeholders, who provided insight into responsibly transforming and implementing technology in the organization.

Garnering these insights, the CDAO identified four areas of tech adoption: enablers, quality data, analytics, and AI. Margie Palmieri, the Deputy Chief Digital and Artificial Intelligence Officer at the CDAO, explained each area during a recent interview with Bloomberg Government, which illuminated how industry can work together with the DoD.

Enablers

As Palmieri noted, the CDAO can execute its mission only by first attracting, recruiting, and retaining workers who are capable of breaking technological barriers, tackling key problems, and driving innovation and growth.

“The first piece [of the hierarchy] is foundational enablers,” Palmieri explained. “We need the right acquisition foundation that allows us to reach vendors in the commercial world that have cutting-edge capabilities.”

To attract foundational enablers, the CDAO launched a pilot program called Defense Digital Corps (DDC), where technologists across the nation can work on emergent issues and grow their careers within the DoD. Similarly, the CDAO is working on streamlining the acquisition process of critical technologies, using decentralized procurement vehicles like Test &Evaluation Blanket Purchase Agreement and TryAI Commercial Solutions Opening.

Quality data

The data that DoD uses for analytics and AI must have seven key qualities, according to Palmieri. She explains that for data to be considered “quality data” it must be VAULTIS: visible, accessible, understandable, linked, trustworthy, interoperable, and secure.

Whereas data lakes or warehouses are often the norm to maintain data quality within businesses, the CDAO uses a data mesh, which unites data sources across joint forces and gives control to experts who manage the data as a product within a decentralized governance framework. Because the DoD is a decentralized enterprise, Palmieri notes how this form of data architecture is consistent with the department’s culture.

To improve data quality on the warfighter side of the organization, the CDAO is leading a series of Global Information Dominance Experiments (GIDE). GIDE looks at data integration across combat commands and helps inform Joint All-Domain Command and Control (JADC2) solutions. The experimentation team is made up of civilians and military personnel who are committed to providing quality data to warfighters and industry to execute DoD’s mission.

Analytics

According to Palmieri, a lot of decision making happens within an analytic window by taking quality data and creating dashboards and other visualizations to better understand how to manage personnel and operations within the DoD.

The CDAO leverages contractors to develop state-of-the-art analytic tools to support decision makers from the strategic level to tactical warfighters. To ensure quality control, Palmieri noted that there is testing, feedback, and refinement, so that the operator knows how to use each tool and understands clear metrics to evaluate overall efficacy.

Artificial intelligence

As CDAO scales, Palmieri emphasized that it is essential that AI is exercised responsibly among DoD divisions, partner organizations, and commercial industry.

“There is a lot packed into the word ‘responsibility’ but it’s important that we understand AI’s capabilities and pros and cons,” she said. “It’s not like a hardware system where it doesn’t change much in the field unless you purposely interfere and make some changes. AI is different; it’s a learning system.”

Palmieri mentioned that people will still need to come in and train or retrain algorithms but that must be done with careful consideration to ensure that AI systems are accurate, secure, and up to date. CDAO is scaling responsible AI development through scaffolding, which is a means to bring together government and industry to work on AI tools, best practices, services, and capabilities to strengthen national security.