It’s impossible to overstate the centrality of data in the AI equation. As Dr. Persons explains, “you’ve got to get the data right. The data are the fancy Latin term sine qua non—‘without which, not.'” In other words, agencies need a steady flow of good, clean, reliable data in order to train the machines. Without a firm commitment to this principle, it’s impossible to achieve the promise of improved efficiency and effectiveness.

“If you don't have that data, the project is much, much more difficult,” adds Measure. “I've worked on some projects where we don't have that nice clean database [and] it can completely prevent you from using the tools in the first place.”

When Financial Systems Branch Chief Gisele Holden helped to launch the National Science Foundation iTRAK program, a multimillion-dollar modernization initiative, this was her starting point. “We cleaned up our data several years before we even purchased a system,” she says. Cleaning up data at the start enabled her to leverage the best of her agency’s data stores, while utilizing only the minimal amount of data needed. This approach proved cost-effective and also helped to make data governance more manageable.

In addition to cleaning existing data repositories, agencies can also look to the AI itself as a source of data. Each customer interaction, each iteration of the AI application, generates new insights into citizen need and agency resources. That data in turn helps to fuel further improvements within these machine-driven processes.

All this will depend on having a robust data management infrastructure, one that reaches across the agency and can support a range of exploits. “It should not be on a project-by-project basis or use case by use case. It really is about building a foundation that outlasts any single endeavor,” says Dr. Mona Siddiqui, Chief Data Officer for the U.S. Department of Health & Human Services.

''If you don't have that data, the project is much, much more difficult. I've worked on some projects where we don't have that nice clean database [and] it can completely prevent you from using the tools in the first place.''

— Alex Measure

Why AI?

With all the recent talk about AI, it’s tempting to see this as just another high-tech trend, a buzzword that will burn brightly but briefly. In fact, the reason we’re talking about AI in such depth is because this is truly a transformative technology.

AI is going to change the way government executes the mission, manages its internal operations and it will profoundly impact the nature of the citizen experience. What will be the nature of this transformation? That’s exactly the question government is tasked to answer.

In order to move forward into an AI-driven era, agencies need to have a fuller understand of the impacts of machine learning, especially when it comes to augmenting the government worker.

For the first time, computers are adapting to support human need, as compared to other tech-driven revolutions in which humans have adapted to the computer. There’s a unique opportunity here to explore the ways in which workers can be more productive, as AI empowers them to deliver more personalized services to constituents and stakeholders.

If that’s true—if we are fundamentally reshaping the relationship between the people and their machines—then it’s imperative to talk about the rules of the road. We need human-centered research around this changing relationship. We need to establish expectations for how people will interact with these systems, and for how the systems will provide feedback.

The pioneers have realized that human-centered design of these new interactions is paramount. That's why we're seeing such concentrated interest not just in AI, but in the human impacts of this transformative technology.

Connect with us

Subscription Center
Stay in the Know with Our Newsletter Stay in the Know with Our Newsletter