Artificial Intelligence is being implemented across government to modernize and automate traditional manual processes. For many organizations, this means taking paper-based, tedious, error-prone tasks and turning them over to a machine for automated completion. Beyond using AI to hand off tasks best completed by machines -- those that are rote and repetitive -- agencies are also looking at ways to introduce the technology into already complex human-driven activities to make them even more effective and efficient.
Researchers at Dartmouth College's Department of Computer Science have taken a technique that proved valuable in WWII and applied AI to extend the usefulness of the method. A canary trap is a technique that plants different instances of false information in documents. If one of those documents is leaked, the canary will "sing," identifying the leaker. For example, in WWII British intelligence agents planted false documents on a corpse to trick Nazi Germany into preparing for an assault on Greece while the Allies invaded Sicily. The team at Dartmouth created a modern version, WE-FORGE, that plants different instances of false information in documents. The process is relatively simple when creating a small number of variations in a handful of documents, but to extend it to large scientific or technical documents, AI is essential. WE-FORGE uses natural language processing to generate multiple fake files that are believable yet incorrect.
The development and use of Artificial Intelligence (AI) became official policy of the United States with the signing of an Executive Order in February. This order outlines and directs America's government-wide push to advance the use of AI through research and public/private partnerships. In the ensuing months, the Department of Energy has emerged as a leader in these efforts.
In September 2019, the DOE initiated the Artificial Intelligence and Technology Office (AITO) to help channel the department's vast resources across its national lab facilities. These efforts are paying off as DOE partners with Health and Human Services and Veterans Affairs as part of the COVID-19 Insights Partnership with the goal to increase data sharing and analysis in the fight against the spread of COVID-19. The DOE is also pressing ahead with private partnerships announcing the First Five Consortium with Microsoft, the Pacific Northwest National Laboratory, and the Defense Department's Joint Artificial Intelligence Center (JAIC). Together they will develop AI-based solutions for data-first responders.
State and local agencies are home to some of the most innovative ideas in government. Their use of artificial intelligence (AI) is no exception. Localities are embracing AI as a way to make sense of all the data they hold to better understand how citizens are using their services and where gaps may exist. A survey from the National Association of State Chief Information Officers (NASCIO) released in the fall of 2019 found that 32% of those surveyed "strongly agreed" that AI and related technologies can help them meet citizen demands and improve operations. Specifically, the survey found that nearly 50% of respondents planned to use AI as a way to shift workers away from rote tasks and toward high-value activities.
Taking a look around the country, we see some interesting applications of AI at the state and local level.
With the government fiscal year starting in October, our Federal government gets a head start on their New Year's resolutions. As we launch into a new year--a new decade, even--we wanted to take a quick look at government technology priorities for 2020 and beyond.
Cybersecurity - In the past decade security has transitioned from a stand-alone technology that had to be added to planning and systems, to a utility-type service that is baked into every piece of technology deployed within government. This fall, Federal CIO, Suzette Kent shared her focus areas for the next year (and beyond) to include cross-agency information sharing, improved identity management, and increased workforce cybersecurity literacy.
Reskilling - The introduction of automation into administrative functions is driving a need for employees to be re-skilled. While machines are not taking over the jobs of humans, they are improving efficiency in many roles, freeing up time for people to take on more complex (and frankly, more interesting and more important) roles within an organization. Continue reading →
Artificial Intelligence (AI) continues to dominate tech headlines. Now, rather than learning what the technology could mean for government, we're reading about where it's being implemented, and the results being achieved. A recent report found that AI is no longer considered optional, but rather a critical component to managing and using large amounts of data. IT leaders in government are looking to AI to automate routine, data-oriented tasks, ease access to diverse sets of data, prioritize tasks based on the benefit to the organization, and generally keep track of ever-growing streams of data.
The Intelligence Community (IC) has long been a top consumer and analyzer of data in government. Not surprisingly, they have embraced AI technology to supplement the work of analysts by reducing the amount of manual data sorting with machine-assisted, high-level cognitive analysis. AI is being used to help triage so the highly-trained analysts can spend their time making sense of the data collected by looking at the most valuable and seemingly connected pieces.
Health and Human Services (HHS) implemented an AI solution when they needed to quickly procure Hazmat suits to meet the response to an Ebola outbreak. Procurement officials were able to use AI to make like-to-like comparisons among products. After the initial tactical analysis, the acquisition teams were able to use the data gathered on department wide pricing and the terms and conditions to better define parameters for ten categories of purchases.
Despite the successful implementations in many agencies, AI is still in the pilot and introductory phase. The Air Force is making it easier to begin experimenting with AI. Because the DoD has strict rules about what can be put on their networks, it is difficult to introduce new technologies into the production environment. The Air Force has created a workaround with the Air Force Cognitive Engine (ACE) software platform, a software ecosystem that can connect core infrastructures that are required for successful AI development (people, algorithms, data, and computational resources).
HHS is looking to use AI to analyze dated regulations as part of their AI for deregulation project. The pilot has found that 85 percent of HHS regulations from before 1990 have not been edited and are most likely obsolete. Using AI to flag regulations with the term "telegram," for example, will begin the prioritization of data that needs to be looked at by humans.