State technology officials say data quality is increasingly important, but they are not actually implementing data quality programs, according to a preview of survey results shared by accounting firm Ernst & Young at a technology conference on Tuesday. He said very few states have done so.
42 of 46 states rank data quality as “important,” “very important,” or “very important,” according to data from a preliminary survey of state technology officials presented by EY executive Chris Estes. did. However, 78% of states surveyed said they did not have a data quality program.
Bill Smith, Alaska's chief information officer who spoke during the session, said the data readiness gap is leaving states scrambling to keep up with the demand for high-quality datasets that has skyrocketed with the advent of generative artificial intelligence. I guessed that this was the result.
“Over the past six to 12 months, this has increased in priority and importance. [of data quality] But that's not enough time to start a program,” Smith said. “That's what we should have been doing all along.”
States are now rushing to establish AI governance policies, working groups, and data frameworks to take advantage of a suite of new generative AI tools proposed by the private sector and incorporated into existing software. Smith said that while his IT organization is among those that do not have a formal data management program in place, some institutions in the state, such as those that handle health data, do have formal data management programs as part of their compliance efforts. He said he operates a data management program.
Christy Burris, chief data officer for North Carolina, who spoke during the session, also said that states will always have low-quality data, but the goal of data governance is that when new potential AI use cases emerge, He said the priority is to get more data to prepare for the future. important project.
Burris cited “high-profile” criminal cases in the mid-2000s as examples of states failing to find perpetrators because state data was spread across many agencies and there was no easy way. He did not specifically mention it. Combine and share. He said trust is a key element to improving data governance within state government, and North Carolina currently operates a data portfolio for health care, “child and family welfare,” and longitudinal data services. It is said that they are doing so.
“Data sharing moves at the speed of trust,” Burris said.
96% of state officials responding to an EY survey agreed that increased adoption of AI and generative AI will “impact” the importance of data management. However, states' ability to enact data and AI programs varies. According to the study, 26 states are “reactive” in terms of data maturity, six states are merely “aware” of their data, and only 12 states are “proactive” and two are “proactive.” It was only the state. The states said use of the data is “controlled.”
“Really meaningful use cases for AI will have to be postponed because the data is not ready,” Smith said. “Or maybe you have a very high-priority use case, and to overcome that hurdle, you create an isolated, bespoke dataset that's curated to that need. But downstream, You have 100 pieces of data that aren't necessarily correlated or updated, so it's a huge burden.”
Burris said her data team is part of an informal AI and data workgroup that has recently demonstrated the importance of collaborating across the various stakeholders surrounding government agencies and state governments. said.
“We can move their use cases into the framework and seriously consider implementation, but we're going to proceed with caution,” she said, adding that her state's first use cases are “low “risk'' and provide generated AI to the general public, adding that they are not at risk.
EY plans to publish the completed study this summer.