AI is already happening in digital government: Real-world examples of practical machine learning
Machine Learning in Digital Government
Machine learning (ML), a subset of artificial intelligence (AI), involves training algorithms with data to the point where they improve themselves independently over time. How can government leaders make use of this technology to further the public interest? Where has ML succeeded in the public sector over the past 20 years?

Thentia is a highly configurable, end-to-end regulatory and licensing solution designed exclusively for regulators, by regulators.

RELATED TOPICS

Thentia is a highly configurable, end-to-end regulatory and licensing solution designed exclusively for regulators, by regulators.

RECOMMENDED FOR YOU

SHARE

Share on linkedin
Share on twitter
Share on email
Share on facebook

What if government leaders were able to provide you with live traffic data that would allow you to choose the safest and most efficient route for your every commute? What if public sector tools could allow educators to figure out which struggling students are most at risk? What if public safety officials could distribute resources with maximum efficiency, reducing ambulance response times across the board? In many ways, this future is already here, and it’s thanks to machine learning.

What is machine learning?

Machine learning (ML) is a process by which software programs can improve their accuracy and ability to predict future trends based on historical data. Government leaders use ML for, among many other things, predicting crime, analyzing public opinion, and detecting fraud. ML is comprised of four basic approaches: supervised learning, unsupervised learning, semi-supervised learning, and reinforcement learning.

  • Supervised learning: Data scientists provide the algorithm with a specific dataset and program it to look for trends that have been specified by human operators.
  • Unsupervised learning: The algorithm is given a dataset but allowed to analyze data on its own terms, identifying patterns not necessarily specified by data scientists.
  • Semi-supervised learning: A combination of the prior two approaches, this can involve telling the algorithm to search for certain patterns but still giving it free reign to reach its own separate conclusions.
  • Reinforcement learning: Data scientists incentivize the algorithm, in a game-like interface, to learn how to make a set of sequential decisions, rewarding it for positive outcomes and penalizing it for negative ones.

With government leaders in possession of a massive wealth of citizen data, they find themselves presented with an opportunity to use ML to make crucial decisions about policy planning, public health, public transit, and much more. So how have governments leveraged this technology in the 21st century to protect the public interest? Where have they succeeded?

How do governments use machine learning?

Detecting fraud

In 2017, the Illinois Department of Innovation and Technology (DoIT) set out to further develop the state’s use of artificial intelligence (AI) and ML algorithms to provide citizen services more efficiently and effectively. Other state departments, like the Illinois Department of Revenue (IDOR), have already begun using the technology. The IDOR’s application of machine learning has mostly been used to find evidence of tax fraud.

By providing ML algorithms with historical data sets containing past incidents of tax fraud, the department has been more easily able to identify cases where individuals need to supply clarification or additional information in their tax filings. According to an article from GCN (formed in the 1980s as Government Computer News), “the model assigns a fraud probability to taxpayers whose returns gets flagged,” making it easier to flag new suspicious cases.

Public education

The state of Illinois has leveraged ML technology for more than just tax purposes. In 2018, for example, the Illinois State Board of Education (ISBOE) used algorithms in attempt to target at-risk students for one-on-one intervention. By supplying the algorithms with data on students who did not complete high school, as well as data regarding their school performance and demographic information, regulators can assign students a level of risk and use this to determine which struggling students need attention most urgently.

Public safety

In 2017, the Fire Department of Cincinnati partnered with ML technology leaders to create a live dispatching system powered by AI. The system was designed to predict which incidents would result in hospital transport and require a medical transport to be dispatched. The algorithm considers historical data regarding medical emergencies as well as weather data, time data, and geographic data.

Using their new system, the fire department was able to reduce delays in getting patients to hospitals by an average of 22 percent. Thanks to ML algorithms, the department was able to better distribute the thin herd of ambulances at their disposal, while also making sure incidents treated on-site received the appropriate level of response. As a result, response times fell across the board.

How can government leaders leverage machine learning in the future?

There are several promising applications of ML being piloted in North America and throughout the world. Among these is the use of historical and real-time traffic data to shorten citizen commute times and improve traffic safety. The city of Denver, for example, has developed the “Denver Go” app, which compiles data from many different sources, concerning many different modes of transportation, to give citizens information on the timeliest, most sustainable, and most affordable routes available to them.

In Los Angeles, public officials released a live map allowing users to determine the safest route and mode of transportation by analyzing data on pedestrian, cyclist, and motorist fatalities. The city of Atlanta released an app to help citizens navigate around a recent major bridge collapse. In Kansas City, the city’s downtown corridor displays live data and visualization of traffic patterns across multiple modes of transportation.

It all stands to show that ML algorithms are part of the future in digital government. By making sure information is collected efficiently and organized effectively into machine-friendly datasets, and by making sure citizen data is protected as these algorithms exert influence on more elements of our daily lives, government leaders can work to ensure the public interest is protected as we move further into our data-driven age.

SHARE

Share on linkedin
Share on twitter
Share on email
Share on facebook
Jordan Milian
Written byJordan Milian
Jordan Milian is a writer covering government regulation and occupational licensing for Ascend, with a professional background in journalism and marketing.

Featured

Ascend Magazine lives at the nexus of regulation, licensing, public policy, and digital government. We share news, insight, and exclusive commentary from leaders in regulation and technology. 

OCCUPATIONAL LICENSING REFORM

VOICES

CYBERSECURITY

LICENSE PORTABILITY