The Partnership for Public Service has issued a report that examines how artificial intelligence (AI) is being used by federal and state authorities in a variety of areas.

The research for the report, “The Future Has Begun: Using Artificial Intelligence to Transform Government,” was performed in collaboration with the IBM Center for The Business of Government and includes four case studies. The first case study involved using AI techniques originally created at the University of South Carolina to fight crime more effectively. The software was originally used to fight domestic terrorism by helping federal and state officials determine potential targets and make recommendations on randomizing patrol routes, security schedules for police officers, boat patrol routes, and assigning air marshals to flights. In a later application of the same software, the same system was given to wildlife rangers in Africa to assist in making decisions on which wildlife areas to patrol on any given day to protect both animals and plants. This was in response to presidential Executive Order 13648 issued in 2013 to have the Agriculture, Treasury and State Departments assist in combating wildlife crime.

A second case study involved using AI at the Labor Department’s Bureau of Labor Statistics to automate tasks of a tedious or repetitive nature, including the manual reading of hundreds of thousands of survey responses about workplace injuries and illnesses each year. This effort is done for the purpose of spotting trends and making recommendations on how to prevent injuries and illnesses. Using machine learning technology to code responses resulted in the computers coding “more accurately, on average, than a trained human coder.” Moreover, employees of the Bureau can now “focus on more complicated cases that require human judgment, shifting from mind numbing to more interesting tasks.”

In a third study, a Kansas county used AI learning in a pilot program to offer preventive services to a vulnerable population. Based on data sources including a local emergency medical services department, the county mental health center and the county’s joint database of law enforcement, court and corrections data, county staff were able to use a machine learning algorithm to predict recidivism. The algorithm identified 200 incarcerated people who were at risk of returning to jail.” The county would track the release date of each individual, and upon their release, contact them to offer county services. The report notes that “officials hope this enables the county to help more citizens and prevent more family crises.”

A fourth case study involved the Department of the Air Force’s plans to use AI to help with the Federal Acquisition Regulation (FAR) process. This includes answering queries from federal contract officials and contractors concerning the FAR, based on the computer’s knowledge of regulations, contract cases, and policy and training materials. According to the Air Force, AI could help contracting officers “quickly find answers to difficult questions so that they can focus on creating agreements and use the flexibility available in the procurement regulations.” It could also open the doors to more innovative services and technologies because it would make the contracting process more accessible to first-time contractors and small businesses. The report notes that any business interested in an Air Force contract would be able to query the software about bidding on a contract and request a list of all the contracts that it is eligible to bid on. The report concludes by noting in part that “it is vital for government to make a strategic investment in understanding how to maximize AI’s benefits and use it to improve agencies and government as a whole.” Notably, however, the report also goes on to caution that:

Questions remain pertaining to information privacy and cybersecurity, AI systems’ trustworthiness and reliability, and the role of these systems in operational and organizational transformation. Other issues to address include the prevention of unwanted bias in AI as well as the use of AI to counteract human bias.

The Partnership for Public Service is a nonpartisan, nonprofit organization based in Washington, D.C., that works to revitalize government service.

A link to the report can be found here and a recording of a program held in connection with the release of the report is available here.