On October 12, the General Services Administration (GSA) held an event called Future Services Now to highlight the ways the federal government is using emerging and innovative new technologies to improve government services. One of the main takeaways from the event, hosted by GSA’s Emerging Citizen Technology Office, was that some of people’s biggest frustrations with the government—inefficiency, bureaucratic sluggishness, non-user-friendly services, and more—could soon be a thing of the past thanks to the use of data and AI.
While the use of automation and AI in the private sector is prevalent, the federal government has struggled to figure out how to take advantage of these technologies, but this is changing. Ed Burrows, who leads robotics process automation (RPA) for GSA, stated that “We didn’t expect a month ago that we could have automated processes that take 1,000 hours with a bot. Now we know that we can.” RPA is the automation of rules-based, typically manual, repeatable processes, which federal agencies are looking to streamline to adhere to the requirements of M-18-23, guidance from the Office of Management and Budget (OMB) directing agencies to shift the focus of federal employees from low-value work to high-value work. The Food and Drug Administration (FDA) has already seen promising results from its experiments with RPA in the Center for Drug Evaluation and Research (CDER), which evaluates applications for new drugs, generic drugs, and biologics. CDER developed a bot to automate this evaluation process, which typically involves nearly 20 full-time workers manually entering data from PDFs into a review system, and was able to improve the average processing time by 60 percent, saving 8,000 hours of manual work. CDER also developed bots to improve the data quality in its repository of drug-sponsor-submitted data, which it uses in the regulatory review processes for drugs but is plagued by messy data that results in delays and reduced performance. These data quality bots has saved 5,490 hours of manual work.
In addition to improving back-end processes, automation can make it substantially easier for the public to interact with government services and access more relevant information faster. KPMG demonstrated an automated service it developed for stopbullying.gov, an initiative from the Department and Health and Human Services designed to serve as a resource for people looking to identify and reduce bullying. KPMG developed a smart chatbot with Google’s TensorFlow machine learning library that makes it easy for people to request information about how to stop bullying, report it to school officials, or get help if they themselves are victims of bullying. The chatbot can integrate with third-party services, such as Facebook Messenger and Amazon’s Alexa, making it easier than ever for people to interface with a government agency. KPMG is developing similar tools for opioids.gov and HIV.gov, which can make it easier for people to access potentially life-saving health information and services.
Though these new tools are promising, the federal government needs to address several key challenges in order to break away from the high-cost, low-satisfaction citizen service model it is stuck in. Large amounts of data silos limit the ability of agencies to share and use information effectively. High turnover, long training times, poor data access, and the overwhelming amount of boring, “low-value” work means customer service agents in federal agencies are often ineffective. And many agencies fail to leverage data and analytics in their citizen service processes to develop insights about how to improve, such as by developing personalized processes or predicting customer intent—an opportunity the private sector recognized long ago.
However, GSA’s event demonstrated that the federal government is serious about overcoming these challenges and has already made encouraging steps to use automation and AI to make government better. As Curtina Smith, OMB acquisition analyst, noted on a panel, this is simply “a matter of ‘when,’ not ‘if.’”