Harnessing the information hidden within the federal government’s truly vast stores of data is one of the biggest challenges in pursuing digital transformation.
Panelists during the Advanced Technology Academic Research Center’s Digital Transformation Virtual Summit on Jan. 18 shared how they are tackling the issue and some of the outcomes they are hoping to achieve.
Scott Simpson, Digital Transformation Lead for the Department of Homeland Security, for instance, envisions using data to innovate the procurement process. He runs the Procurement Innovation Lab for DHS.
“I’m an innovation coach,” he said. “I had an idea to innovate procurement [by teaching procurement officers to] apply agile methods to help the team across all DHS components to buy faster, smarter and more efficiently.” Currently, one idea being considered is how to use artificial intelligence to more accurately assess past performance, one of the criteria for most government solicitations.
Ron Thompson, Agency Chief Data Officer and Deputy Digital Transformation Officer at NASA, is wrestling with the changing nature and value of the data.
“This is beyond tagging and classification,” he said. “We are holistically, across the enterprise, in our infancy of knowing what’s in the enterprise. [If] you look over time, the sensitivity, the classification, changes … We have to look at a use case and see how we’re using the data, make sure we have intelligent data management in place [so] that it might not be tagged or classified at that level over time.”
Kenneth Clark, Assistant Director, U.S. Immigration and Customs Enforcement, said finding the right data is getting harder and harder as the volume of data keeps exploding.
“It’s important to have the tools, artificial intelligence or machine learning, to help you sift through this massive amount of data,” so the agency can use it while still protecting it. It’s about finding “the right balance of what information you need, where you’re going to get it, and how you’re going to protect it,” he said.
The human factor must be considered, as well, all three agreed. Most users have no awareness of the long chain of datasets and IT systems that provide the framework for the user endpoint, nor do they necessarily recognize that the data they enter is part of that chain. Asking them to do something that seems simple, like changing the order of fields on a screen where they enter data, leaves them frustrated.
“So we try to show them what AI and all these [tools] mean to them,” Simpson said. “There are millions of records for hundreds of thousands of vendors … Add AI and you can really reap the benefits.”
Then users may say, hey, I can see what I get out of this, and they may take a few minutes and put more data in correctly, he added.
“You want to use what you’ve learned in the past and make things better for the future,” Clark said. “When you look at some of the technologies out there, AI, for instance, one of the threats is having bias in the data … [So we ask] why do we have it structured this way, why did it fail? We don’t want it to fail again when it’s used in AI.”
“It’s that linkage to mission and mission outcomes,” Thompson said. “When people hear ‘data strategy’ they think of records management. [Having] a chief data officer is new to the federal government … NASA is taking what we call a light, steady hand to our approach. [The] NASA culture is engineering. It takes time [and] consistent pressure.”