COVID-19 is driving greater adoption of machine learning and automation in the back-office. How are agencies using these powerful tools to meet the mission today, and what guidance do the front-runners of adoption have to share?
It’s no secret that government has “turbo-charged” (to borrow a phrase from Wharton’s May article on innovation) its adoption of technologies that enable the delivery of data-driven insights faster as a result of the pandemic.
When viewed against the backdrop of the continuous struggle to deliver on the mission despite the hindrances of legacy IT systems and archaic business processes, the rapid innovation that has occurred within government since March is nothing short of extraordinary.
Automation and machine learning have taken greater hold as mechanisms for putting critical data at the fingertips of decisionmakers across agencies.
At ATARC’s recent event focused on automation and machine learning for better citizen service, several agency leaders weighed in on how their organizations are implementing these technologies, and how they envision leveraging them in the future.
The U.S. Small Business Administration, which is responsible for the Paycheck Protection Program, issued more loans in the first three to four months of the pandemic than the total number of loans ever issued in its 67-year history (10 million loans in the amount of $750B). This, while moving their workforce to 99 percent remote.
Sanjay Gupta, the Administration’s Chief Technology Officer, described how the agency quickly created more email storage thanks to the modern, cloud-based foundation already in place—the agency was overwhelmed with email even before the CARES act was issued—and began to use artificial intelligence (AI) and machine learning to sort and label the deluge of inbound email.
The National Technical Information Service within the Department of Commerce, charged with providing innovative data services to agencies across government, shifted its focus to helping agencies meet the challenges arising from the crises when the pandemic hit.
Chakib Chraibi, NTIS’ Chief Data Scientist, described how his teams are working with the Department of Health and Human Services (HHS) to use anomaly-based machine learning to help prevent Medicaid fraud, and with the Department of Homeland Security (DHS) to both model the impacts of catastrophic events to our national critical infrastructure (pre-pandemic) and examine how COVID-19 is impacting critical infrastructure sectors today.
Importantly, Kathy McNeill, GSA’s Director of Artificial Intelligence within the Federal Acquisition Service’s Centers for Excellence, Technology Transformation Services, described the increased demand for robotic process automation (RPA) as a way to drive adoption on low-risk, low-level work to free up time for knowledge workers.
What advice did the speakers have for agencies as they continue to bring AI, machine learning, and automation into their back-office processes?
From Ms. McNeill: Look at the data first. Understand its sources, and leverage some of the new tools out there today to start to sift through and help understand what state it’s in. Don’t try to get the data perfectly clean first. Machine learning is iterative and builds on itself, so it can’t get off the ground without some data input to begin the process. The Centers for Excellence can help.
From Dr. Chraibi: Start early. AI, machine learning, and automation tools run on data, and data is created every day.
And, from Mr. Gupta: Take a look at the SaaS tools you are already using. Many have machine learning built in, so you may be using the technology without even knowing it. Looking at what those tools do for you can help spark new thinking about other applications.
No doubt the explosion of innovation in government will continue, and with it, greater adoption of AI, automation, and machine learning. How agencies will leverage these powerful technologies to serve the people moving forward is anybody’s guess.
I can’t wait to find out.