There's No ‘Easy Button’ for Making the Most Out of Data

Good data is the key to so many of the modern services that state government provides. It’s often the foundation for public policy decisions, critical health and human services, and so much more.

But behind the quest to mine the most useful data lies a substructure of other considerations. State and industry leaders discussed the pressing issues surrounding data use during the California Government Innovation Summit* on Aug. 20, offering their perspectives on the necessary guardrails, potential security risks, and future innovations.

A primary concern for Ryan Estrellado, director of data programs with the relatively young California Cradle-to-Career Data System, is building a relationship with users around data they can trust. The department was funded by the Legislature in the 2021-22 budget to “provide data-informed tools to help students reach their college and career goals and deliver information on education and workforce outcomes,” according to its website.

“What we think about a lot is the question of, how do you build trust with the California public? And with something like a massive data system, and one way that I think about that is it’s not so different from building trust with people in your everyday relationships; it’s about having repeated everyday interactions that are honest, authentic, transparent. So that’s something that we try to implement through our governance,” Estrellado said.

The same could be said for the approach to scaling user-centered design, where Estrellado advocated building UX requirements into the broader governance model and the culture of the organization.

“I think lesson No. 1 is [that] if you care about user-centered design or any particular value that you choose as an organization, work it into your governance and make it a part of your everyday process,” he said.

Office of Data and Innovation (ODI) Chief Data Officer Jason Lally agreed, adding that people and organizational culture are critical components of successful data work — from building out equity efforts in a meaningful way to removing the myriad barriers to innovative data use.

“What we want to do, I think my role is, is to really identify what’s getting in the way of the humans in the data, and the humans of the data to do the great work they already are doing and want to do, and really prioritizing initiatives around that,” Lally said. “So it’s really supporting the humans in all of this, that’s going to be where we focus.”

John Ohanian, chief data officer for the California Health and Human Services Agency(CalHHS), said much of the technology necessary for sharing data like longitudinal health records technology that his agency and others rely on is already in place — the key is providing adequate access to the right users.

“And so going into looking at our strategy here at CalHHS, and kind of what we see as the potential is we know that most information is digital; if not, we know how to make it digital, and we know that the health information is starting to exchange, but what we don’t have is it happening globally and wide scale,” Ohanian said.

As AI and generative AI (GenAI) continue to seep into state government functions, Ohanian stressed that this is the time to make preparations for the evolving technology — not so much with rushing the most novel use cases, but by laying a more solid footing for the technology.

“What I’m trying to stress to our data community is whatever direction GenAI takes and whatever runway it’s on, our work today is to double down, triple down on making sure our data is correct, making sure we have good governance, making sure we have good standards …,” he said.

It should come as no surprise that Google, the company so famous for its search engine, sees the need to make data more discoverable while it’s moving through data systems. This sort of innovation will be a necessary stepping stone for how agencies engage with GenAI systems down the line, said Charles Elliott, field CTO and practice lead with Google.

“It certainly looks like an easy button in some use cases, but it’s not. The better kind of data quality and controls you have in place, the better those models tend to serve you,” Elliott told attendees.

While putting scrubbed data in an accessible location could be counted as a win for many organizations grappling with the modern flood of structured and unstructured data, the next step forward becomes retrieving it on demand, said Jumbi Edulbehram, NVIDIA’s director of global business development.

“Think about being able to go through massive amounts of multimodal data, giving you an answer in less than a second with the context of what you want to know, and then generating the answer in any language,” he said.

*The California Government Innovation Summit was presented by Government Technology, sister publication of Industry Insider — California.