The rapid adoption and scrutiny around ChatGPT, the latest artificial intelligence model, highlights the exploding potential of AI. It also underscores a stark delta between AI’s current capabilities and human willingness to embrace them.
This untapped potential is perhaps most apparent in the public sector where AI is both championed and feared - offering exciting federal use cases but lacking policy guidance and maturity at scale.
So how can agencies better leverage AI? Results of a recent SAIC survey uncover a need to improve overall AI readiness to enable the technology’s integration and benefit from it.
The independent survey of federal government executives revealed that while AI is on respondents’ radar, fewer than one in five are “very” likely to adopt AI in the next year. Further, it indicates that AI readiness is a major barrier to implementation with one-third of respondents stating they do not believe their agency is ready for AI.
To overcome this paradigm, agency leaders must clear up the common misconception that AI/ML infrastructure, data governance and efficiency must be perfectly aligned to get started with adoption. In reality, leaders can pilot AI confidently while developing equally important governance processes and risk frameworks in parallel.
A Data-First Posture
Good data management opens the door for successful AI, and the Pentagon’s new Chief Digital and Artificial Intelligence Office, ranks driving high-quality data among its top priorities.
The CDAO recently reassembled a Global Information Dominance experiment called GIDE V to underscore how “data, analytics and AI can improve joint workflows in a variety of missions” including its signature Joint All-Domain Command and Control, or JADC2, effort.
By connecting AI to its ability to automate repetitive processes and bring intelligence to many tasks at speed, the Pentagon is showcasing its ability to support the warfighter, thus building confidence in AI-enabled systems.
Agency-Specific Guidance
The federal government has sought to advance AI technology development and adoption through several important bills and resolutions, but agencies need tangible agency-specific guidance and use cases to implement AI at scale.
SAIC’s study revealed that while some agencies have AI-specific guidelines, most have yet to implement a policy with 57% reporting the biggest obstacle to implementation, other than readiness, is policy and governance.
This need for better policy and governance is already being addressed by the CDAO office, which is charged with guiding AI innovation across the department. Additionally, the U.S. was the first military in the world to formally adopt AI Ethical Principles in February 2020 and has an AI Education Strategy in place at all levels to drive change management and AI receptivity.
Optimising Cloud
Improving the cloud environment opens the door for AI and ML by enabling faster, cost-effective government innovation and capability.
Survey respondents believe that adopting more than one cloud allows agencies to take advantage of the best solutions from cloud providers, ensuring they have the right tool for the right job.
In fact, 70% of survey respondents reported that they use two or more clouds.
The top three are Microsoft Azure, Amazon Web Services (AWS) and Google Cloud (not surprising, as commonly used collaboration applications Office365 and GSuite are already included in Microsoft and Google’s clouds, respectively).
DoD’s Joint Warfighting Cloud Capability supports these findings and the Pentagon’s AI and data acceleration initiative by enabling commercial cloud across multiple contracts, providing access at all classification levels and powering the tactical edge.
Enabling Zero Trust
According to the study, agencies see value in DevSecOps, but 39% say they lack the resources and technical skills to manage it.
Ensuring these frameworks and tools are in place is paramount when it comes to addressing AI fears and avoiding disruption, a factor DoD respondents are still concerned with.
Fifty-four percent of DOD personnel were concerned about AI disruptions when compared to civilian employees (39%), likely because the Pentagon is becoming more sophisticated in its AI journey.
DoD’s chief information officer set a goal to fully implement zero trust by FY27, and DOD’s zero trust strategy is helping define the approach at the enterprise level.
Combined, these measures embrace zero trust and DevSecOps to support secure modernization at speed, cloud at the edge and real-time data interoperability.
