Dimitri Kusnezov, deputy undersecretary for artificial intelligence (AI) and technology at the Department of Energy (DoE), revealed that the department relies on data workflows to make high-risk and mission critical decisions.
“As we gather data from so many different places, we have the sense that there is a lot more opportunity out there if we could simply see with more clarity through the complexities of the data,” he said at the AFCEA Bethesda March 19 breakfast. “The challenge is the data situation is simply growing.”
To manage growing datasets and apply them to emerging technologies such as AI and machine learning, Kusnezov and DoE rely on workflows. Kusnezov defined workflows as teams of people that cover a breadth of different skills. By asking different questions and bringing different skills together to solve mission critical problems, the data “can be labeled appropriately, structured appropriately, but fluidly develop into a workflow.”
He compared workflows to a business model centered around the outcomes. When everyone is sitting at the same table, he said, each stakeholder brings skillsets and knowledge that help the problem-solving team “drive outcomes more strongly” through continuous interaction.
“Don’t just say we’re going to have a data strategy,” Kusnezov recommended. “Put it in the context of how it will be used, and the teams that will be built around it.”
Data, however, is uncharted territory. As Kusnezov noted, many agency officials do not realize how useful their data is until they see the outcomes. Then, once officials realize the value of their data, they become hesitant to continue sharing it. The social dynamic this creates is counterproductive to developing effective workflows.
“Creating workflows and archiving data in the right way and models, it’s going to be something where best practices and lessons learned have to be shared. I think it’s a necessary journey we’re all on,” he said. “This is our future, it’ll require us working together, and most of this future is uncharted.”