While artificial intelligence as a concept has existed for decades, we’ve now reached the point where people from all walks of life have seen exactly what this powerful technology can accomplish. Now that they know how AI can make their lives easier, consumers are more likely to notice when an organization isn’t using AI to offer a better user experience. In the federal government, agencies are being called to optimize customer experience in the same way great consumer brands already have.
Before AI became the industry buzz word that it is today, the National Institute of Standards and Technology (NIST) was already preparing for this digital revolution. NIST developed the U.S. Leadership in AI: A Plan for Federal Engagement in Developing Technical Standards and Related Tools[1] in 2019 to set the expectations and priorities of AI adoption as they applied to federal agencies’ missions. The goal was to set technical standards for AI that emphasized data security, enabled innovation, and promoted confidence in the adoption of tools and systems that relied on AI technology.
Fast forward to 2023: Many agencies have included AI in their modernization efforts, either developing their own systems or researching all the necessary components of an AI strategy to ensure they have the appropriate knowledge, partnerships, and standards in place to transform.
As we all know, though, transformation does not happen overnight. As agencies are currently focused on cloud adoption measures, data center consolidation efforts, cybersecurity initiatives and more, their “customers”—or citizens who rely on federal services on a daily basis—will expect to see greater efficiencies in government services develop at a faster pace.
AI can help improve the customer experience by making services more efficient and consistent, so it is critical that agencies future-proof their transformation executions with an infrastructure that has access to the appropriate data sources, can support data surges as well as increased data usage over time, and optimize performance and privacy. Agencies who do not prepare for AI adoption, or who do not fully understand the relevant components of AI adoption, will struggle to satisfy the needs of the American public.
Define your AI strategy now
As AI adoption increasingly moves into the mainstream, federal agencies face many of the same opportunities and challenges as their commercial counterparts. However, there will also be certain concerns that are unique to the government space.
As the White House has just released an Artificial Intelligence executive order[2] addressing security priorities and standards on AI systems, agencies can’t afford to wait for any outside body to force their hand toward adoption. They need a long-term AI strategy, and they need to start building it now. This strategy should clearly define:
- Exactly what they hope to accomplish by using AI
- What infrastructure they need to do AI right
- Which partners they need to work with to make that happen
What can agencies achieve with AI?
Like so many other IT modernization initiatives, an AI adoption strategy for federal agencies must start with the right probing questions:
- Why is the agency planning to adopt AI in the first place?
- What improvements can be made regarding citizen experience?
- How does this tie back to the agency mission?
Another thing to consider is that AI can certainly help agencies achieve great things, but it’s not the magical force it has sometimes been made out to be. Just because one is adopting AI doesn’t guarantee remarkable results. It is important to look for specific opportunities within everyday operations where applying AI could help increase efficiency, eliminate waste, and serve citizens better. For example:
- Federal law enforcement can use AI to quickly analyze evidence, store it securely and share it with partner agencies as needed. This can help speed up investigations and enable faster threat response, thus keeping citizens safe.
- Airport security checks could use AI to quickly verify identities, including cross-referencing photos against no-fly lists. This would help streamline the travel experience for legitimate passengers, without placing them at risk.
- Agencies that operate in federal healthcare can use AI to intake new patients, process their claims and coordinate with care providers to get those patients into effective care plans quicker.
The truth is, anything an agency does that relies on time-consuming manual tasks is a suitable candidate to be modernized using AI. The challenge for agencies is to identify these opportunities within their own processes and then figure out how to capitalize.
What does it take for federal agencies to do AI right?
Just like private businesses, federal agencies need to carefully consider what compute, storage and networking infrastructure they need to enable their AI efforts.
At the heart of any effective AI strategy is data. To get the right results from their AI models, federal agencies need to take in the right data—either from internal sources or by sharing with ecosystem partners—and then process it quickly. The AI workflow centers around an iterative process of model inference and model training, with each step of the process having different infrastructure requirements.
AI requires distributed infrastructure
Specifically, model inference requires very low latency, which means that it typically uses compute resources at the digital edge, in close proximity to data sources. In contrast, model training requires access to extremely high compute capacity. Therefore, it’s more often done in public clouds or in large core data centers.
With different AI workloads best performed in various locations, AI infrastructure is inherently distributed in nature. Agencies need to account for this in their AI strategy. They need access to right-sized compute and storage infrastructure in all the right locations, and they need reliable networking infrastructure to tie the different pieces together.
AI presents unique challenges for federal agencies
On top of everything else, federal agencies face requirements around security, data privacy and sovereignty, and sustainability impact that could be even more onerous than those faced by private businesses. To summarize:
- Agencies need to be able to share and use data, including personally identifiable information, in a responsible manner that does not open that data to unauthorized access or place citizens at risk.
- Agencies must be able to do AI anywhere their mission takes them. This could mean they need to store and process data outside the U.S., while accounting for all the unique cybersecurity and data sovereignty challenges that could entail.
- Agencies need to be able to do AI in an efficient and sustainable manner, taking advantage of sustainability innovations such as liquid cooling to adapt to the high-density nature of AI workloads.
One could also argue that privacy concerns are the primary reason many federal agencies have been slow to start capitalizing on the potential of AI. The need for privacy often leads agencies to silo themselves off, both from mission partners that could share valuable data and service providers that could help deploy the infrastructure to collect, move and process that data. As the following graphic shows, siloed agencies will inevitably find it difficult to succeed with AI.
Any way you look at it, agencies have their work cut out for them. That’s why working with the right digital infrastructure partner is so essential.
Equinix is the AI partner federal agencies need
At Equinix, we’re uniquely positioned to help federal agencies overcome the challenges they face around AI adoption, for the following reasons:
- Global reach: We operate 250 Equinix IBX® data centers in 70+ metros across six continents. This means we’re able to offer a consistent experience and quick expansion opportunities everywhere an agency may need distributed AI infrastructure.
- Access to ecosystem partners: Thousands of clouds, network providers and enterprises are part of the Equinix digital ecosystem. If your AI strategy includes working with cloud hyperscalers, Equinix offers more low-latency cloud on-ramps than any other digital infrastructure company. In addition, Equinix offers a secure, neutral platform to help agencies break through their silos and share data with one another securely, thus supporting better AI models and results.
- On-demand digital services: Our portfolio of digital infrastructure services makes it quick and easy to get the compute, storage and network resources you need, where you need them. This includes Equinix Fabric® for virtual private interconnection and Equinix Metal® for single-tenant Bare Metal as a Service.
- Support for private AI: To reduce cybersecurity and data privacy challenges, many federal agencies will seek to minimize their reliance on public cloud services for their AI strategy. Equinix can support a private AI approach, helping agencies build or access the services they need while maintaining full control over their data and other resources.
To learn more about how Equinix can help you design and execute your federal AI strategy, schedule a Digital Strategy Briefing today. You’ll meet with an Equinix expert to discuss your agency’s unique AI opportunities and then formulate a plan to achieve them.
[1] U.S. Leadership in AI: A Plan for Federal Engagement in Developing Technical Standards and Related Tools, National Institute of Standards and Technology, August 9, 2019.
[2] FACT SHEET: President Biden Issues Executive Order on Safe, Secure, and Trustworthy Artificial Intelligence, The White House, October 30, 2023.
Subscribe to the Equinix Blog