Elements of GSA’s AI Guide to Government

As the age of artificial intelligence continues to rapidly expand and evolve, the General Services Administration is adding to the plethora of input to federal agencies on how to adopt the transformative technology. As contractors navigate the federal AI marketplace, it is important to stay updated on the resources provided to agencies to anticipate areas of need, focus, and compliance.

In a living document titled the AI Guide to Government, GSA provides direction on federal AI implementation to senior leaders in technology, finance, and procurement. Specifically, the document is “intended to help government decision makers clearly see what AI means for their agencies and how to invest and build AI capabilities.”

The guide is divided into seven chapters, with the first chapter addressing the definitions and core purposes for AI use in government. In particular, the document underscores the exponential growth of data and the importance of AI throughout federal agencies to process and analyze inputs while maintaining existing resources.

Key components of the remaining chapters are below.

Structure an AI-welcoming organization.

• Invest in AI within the business centers most able to use it
• Allow AI practitioners to report to mission center and program office leaders for project-based AI vs. to an agency IT office
• Develop a central AI technical resource of tools and security, legal and acquisition expertise (i.e. Integrated Agency Team) to support AI professionals in program offices
• Ensure IT shops provide access to the tools needed for AI development

Address Responsible and Trustworthy AI.

• Draw principles from DOD’s Ethical Principles of AI and the AI EO
• Prioritize diversity, equity, inclusion, and accessibility (DEIA) in AI teams, design, development, deployment, and monitoring
• Ask questions of trustworthiness and reliability often and repeatedly
• Remain updated on AI trustworthiness literature

Cultivate an AI Workforce.

• Identify AI talent within the workforce who exhibit certain qualities (i.e. supports work with data, has a high regard for tech and follows latest tech trend, expresses interest in computer programming, etc.)
• Engage in private sector partnerships to fill gaps in the workforce, augment existing AI talent, in niche AI use cases, or to test potential benefits of an AI solution prior to investment
• Ensure AI retention of talent by keeping AI work closely tied to agency mission
• Develop skills through training, formal education, and attending conferences and exchanges with industry and academia

Build the Foundation for AI Capabilities.

• Evaluate development environments, infrastructure, data management, data manipulation and visualization, and computing power technologies
• Build a data governance organization, to include functions such as a data governance steering committee, advisory group, and/or communities of practice

Mature AI Use.

• Utilize the AI Capability Maturity Model (AI CMM) from GSA’s Centers of Excellence as a framework to evaluate organizational and operational maturity levels

Apply AI to Solve Business Challenges.

• Understand the AI lifecycle and use an agile approach for continuous refresh of the model
• Create a collection of AI use cases in areas rich with data and focused on agency mission that business units can choose from
• Utilize vendor support as needed for scaling an AI pilot to production
• Pursue software tools that use AI as part of another product or service, commercial solutions to augment AI practitioners’ effectiveness, and open source software

Elsewhere at GSA, the agency continues to push government-wide adoption of AI for agency missions. In a recent Call for AI Proposals issued by the Technology Modernization Fund (TMF), GSA encouraged federal agencies to identify AI projects for possible TMF investment.  In accordance with the AI EO, the TMF is prioritizing funding for AI mission-enabling projects with outcomes ranging from enhanced user experience, to improved business operations and data-informed decision-making. For example, the TMF board will expedite qualified project proposals of $6M or less, a project timeline of 1.5 years or less, and if the agency has performed user research or proof of concept research. AI proposals with other cost, project timelines and phases are also eligible for alternative paths of application for funds.