Federal IT leaders want to see their agencies achieve enterprise-wide artificial intelligence (AI) proficiency in the next three to four years, according to a recent survey from MeriTalk underwritten by Dell Technologies and NVIDIA, “From Pilots to Proficiency: Operationalizing Federal AI.”
Almost every week, new policy directives, legislation, or analysis supporting increasing U.S. development of AI capabilities is published, pushing the three- to four-year goal closer to reality.
In recent weeks, efforts to bolster U.S. leadership in AI include:
- White House creation of the National Artificial Intelligence Research Resource Task Force, a group of 12 members from academia, government, and industry led by officials from the White House Office of Science and Technology Policy and the National Science Foundation. The task force will draft a strategy designed to provide researchers with access to data stores, computing power, and other resources needed to drive future innovation;
- The White House Office of Science and Technology Policy launch of AI.gov, a clearinghouse for information on Federal government activities advancing the design, development, and responsible use of trustworthy AI;
- The Department of Defense release of a roadmap for responsible AI implementation;
- Senate approval of the bipartisan U.S. Innovation and Competition Act (USICA), which calls for investing $250 billion in technology research and development. It would create a National Science Foundation (NSF) division focusing on cutting-edge technologies that are critical to the U.S.-China competition, including AI;
- House approval on June 28 of its bipartisan version of the Senate USICA legislation, which seeks to increase Federal spending on scientific research of many types. It would create a new NSF office addressing broad issues, including global competitiveness and national security.
Mindset Shift – Not Just Money – Is Needed for AI Dominance
In March, the National Security Commission on Artificial Intelligence warned that in order to cement the United States’ leadership in AI, significant investment is required – perhaps $200 billion or more over the next 10 years – as well as significant changes in business practices, organizational cultures, and mindsets, led by the White House, cabinet, and Congress.
At the agency level, “executive support is very important – both on the business and IT leadership level. AI is going to represent significant change to an organization over time. Understanding that upfront and having people dedicated to help with that process is key,” noted Larry Brown, solutions architect manager for public sector at NVIDIA.
The majority of the 150 Federal IT decision makers responding to the “Pilots to Proficiency” survey said their agency has more than 10 AI pilot programs. Most also said their agency struggles to incorporate localized AI pilots into overall IT operations.
The most successful AI projects have a very tight scope and a clearly defined set of inputs, outputs, and success metrics, Brown advised. When planning a pilot and its future expansion, teams should include input from partners, infrastructure specialists, application software support teams, and data science experts, he said.
Krista Kinnard, branch chief of emerging technology at the Department of Labor, said her team endeavors to create “[AI] pieces that are reusable and reachable by a wide audience.”
“That’s how we’re thinking strategically about how this technology can really enable our organization,” she said, “but also making it accessible to a wider audience to … solve agency-wide problems.”
A New View of Compute Infrastructure Is Required
Adequate computing infrastructure is another top challenge for agencies pursuing AI innovation, according to the “Pilots to Proficiency” survey. Eighty-one percent of respondents said their agencies need help understanding what an AI-ready compute infrastructure looks like.
NVIDIA, with Dell Technologies, is working with the U.S. Postal Service (USPS) on multiple AI initiatives. One of the first was the creation of an AI infrastructure solution that enables accurate image comparison in order to identify dangerous packages. With the new system, the USPS can complete searches in hours versus days.
The backbone of the system is a graphics processing unit cluster that enables USPS developers to create algorithms for a variety of purposes, which they push to 400 post office locations around the continental United States. The centrally managed cluster allows USPS to switch seamlessly to newer software versions.
“We typically see a requirement for new infrastructure capabilities that span beyond traditional computing platforms,” Brown noted. “It is not just about the application side of the equation. If engineers and data scientists are using laptops from 10 years ago, they do not have the right infrastructure. If an organization has centralized computing resources, but the resources are a few years old and designed to run web servers, that is not going to work.”
