Government agencies face growing pressure to adopt AI, yet their data often still lives in older, separate systems where quality, ownership and definitions vary. That makes foundational data management the precondition for any AI effort the public can trust.

In the concluding installment of a two-part series, State of Texas Chief Data Officer (CDO) Neil Cooke continues the conversation with Dr. Adita Karkera, Chief Data Officer, Government and Public at Deloitte, on how the state is moving its data strategy into practical execution.

For the first time this year, state agencies must submit data maturity assessments to the Texas Department of Information Resources (DIR). Cooke plans to use them to decide where support goes first, especially for small and medium-sized agencies that may lack dedicated data management resources. A second opportunity is closer partnership with DIR’s newer AI and Innovation division, including support for the DIR AI Lab so agency use cases are built on sound data practices from the start.

On AI readiness, Cooke describes records that are incomplete, duplicated or defined differently across programs, unclear ownership, and teams split between moving quickly and moving cautiously. The response starts with a mission need and brings program owners, privacy, security and legal teams in early.

Across the conversation, three ideas carry through:

• Readiness runs both ways: Better data supports better AI, and AI can also help agencies find gaps, duplicates and inconsistencies faster.
• Literacy tied to real jobs: Role-based training grounded in an agency’s own examples and reinforced by named data owners builds a data-savvy workforce.
• Influence without ownership: A government CDO has to create statewide consistency while respecting each agency’s mission, systems and maturity.

In a closing rapid-fire round, Cooke calls the public sector data landscape “maturing” and pushes back on the myth that collaboration slows government work, arguing that it makes the work more durable. Cooke would also retire “AI-powered,” preferring to hear what problem a tool solves, what data it uses and what guardrails are in place.

Did you miss the previous episode?

On the move? Take this conversation with you. If you don’t have time to finish the video, you can stream the complete, uninterrupted audio edition of this episode on the go.