The AI Future of the U.S. Federal Government

The rise of the Chief AI Officer (CAIO) position within the federal government, in light of the directive outlined in the Office of Management and budget (OMB) memorandum on March 28, 2024, has signaled the next phase for federal agencies to advance AI innovation and governance while managing the risks associated with the use of AI. Agencies are now tasked with improving their ability to harness AI in ways that benefit the public and increase mission effectiveness.

Government agencies have put hundreds of CAIOs in place over the last few months. It’s “go time” for AI in the U.S. government. Under pressure, each agency must publish its AI strategy within a year of the OMB memorandum. This doesn’t give much time for something that is so fundamentally transformational for government. Government employees invariably need to cut through the enormous amount of information about AI and related technologies – and do it fast. 

When the Chief AI Officer of the Department of Defense (DoD) was sworn in in April, the DoD issued a statement announcing that ‘the office is responsible for accelerating the DoD’s adoption of data, analytics and AI.” Emphasis on “accelerating”! DoD’s CAIO, Radha Plumb, expressed high aspirations when she stated that her goal was to “ensure we deliver digital and AI technological advancement at speed and scale.” 

Cross-agency collaboration is necessary to enable the sharing and reuse of AI models, code and data. It will require strong leadership to cultivate a shared environment to strengthen and, as the OMB says, “democratize access” to computing, data and storage resources for AI innovation. 

One of the “must-haves” that OMB calls out is that, for AI to help modernize agency operations and improve government services, agencies must remove internal barriers and increase their capacity to adopt AI. Part of this capacity should be focused towards building on the enterprise data infrastructure that will support it. Here’s the key point: you need a plan for developing sufficient enterprise capacity for AI, which will require significantly more storage for data overtime but in the short-term can leverage existing petabyte-scale solutions. From a data management perspective, it will be essential to not only have an adequate infrastructure, but also access to enough capacity to properly share, curate and govern data, which will be used in training, testing and operating AI. This will involve both internally held data and data managed by third parties, such as cloud providers. 

Being able to have access to all this data quickly, easily and reliably will be absolutely crucial as new AI applications come online and it is scaled. 

What will be helpful to accelerate this progress is more intensive dialogue between government and industry partners about what it will take to build this capacity, including data infrastructure, in support of responsibly developing, testing, procuring and integrating AI applications across the federal government. 

Agencies can benefit from closer working relationships with technology solution providers. The National AI Research Pilot (NAIRR) has already taken steps to create public-private partnerships to build a shared research infrastructure. Much more is needed to expand an AI-appropriate enterprise infrastructure, which will have many parts to it, including cyber resilient, automated storage.

The excitement about AI continues to grow. Whether it’s constituent support with AI chatbots, or policy draft generation with ChatGPT, or writing financial narratives with AI budgeting software, or whatever the AI-driven application, the possibilities seem endless. However, decades of technological progress have revealed “lessons learned” that may be worth considering on this new journey of government AI. Examples include:

• Obtain quick wins by taking action on the “low-hanging fruit” by starting with less complicated deployments of AI.
• Apply change management principles, including openness to concerns and objections on a path to alignment and compliance.
• Make time-to-value one of your core tenets of your AI technology strategy.
• The answer may not always be to throw money at the problem; it may require a different way of thinking.
• Adopt a continuous learning approach
• Get as many different perspectives as you can to broaden your “lens.”
• Map out as many “what if” scenarios as you can, as part of your strategic planning.
• Always consider how the new technology deployment will impact people.
• Think long term.

CAIOs and their teams will need to lead efforts to rethink virtually everything to which AI can be applied – systems, processes, workflows, workforce management, and integrations. They will need to consider the implications of safety-impacting AI and rights-impacting AI. 

Government agencies and their partners can tap technology companies for expertise. Partnering with technology companies is a viable aspect of an AI strategy that needs to combine the existing infrastructure with what is to come. 

Our team at Infinidat Federal are having productive conversations with government agencies and departments about planning for capacity expansion in the enterprise-grade infrastructure for the government. If we can share our insights about storage or answer your questions what how storage will need to be revamped to support innovation, please feel to contact me at [email protected].