Data Is the Heart of AI

In the public sector, challenges persist with data sourcing, sharing, training and determining the best ways to use AI. As a result, government agencies continue to tread cautiously while still establishing use cases.

The Office of Naval Research is no exception, and Director and Chief Analytics Officer James Valverde Jr. acknowledged it’s impossible to talk about AI without talking about data, because that’s what informs the models. He added that “a crucial but challenging aspect of that, though, is identifying where the data comes from — for instance, internal vs. external data sources and authoritative vs. certified — and sharing it to work together.”

“People are protective about their data,” Valverde said. “They view data in zero-sum terms.”

To overcome this, the Naval Surface Warfare Center Dahlgren Division focuses on AI’s tactical aspects, viewing it as a form of intelligence, said Harry Dreany, Intelligent Automation Division (H60), Naval Surface Warfare Center Dahlgren Division.

“Especially with the growth of generative AI (GenAI), which we want to embrace … if you don’t
have the right information to feed that capability, then all of a sudden, we put ourselves at a disadvantage,” Dreany said.

Still, the Navy, like many government entities, must be careful because the data it needs for AI models has different classification levels. “If our data is on the edge and it is seen, which it will be, it will be targeted,” he said. “We’re really at a difficult crossroads about how much we embrace the data that we have on the edge, trying to use this technology.”

For more takeaways from a recent discussion about overcoming AI struggles in the public sector, download this resource, How to Use Data, Training to Maximize AI.