‘AI Is a Team Sport’: Georgia Upskills for AI, Tests Agents

Why It Matters

Georgia's team‑based AI strategy demonstrates how coordinated governance can unlock AI value while mitigating risk, setting a precedent for other governments and enterprises.

Key Takeaways

• AI governance requires cross‑departmental responsibility across state agencies.
• Georgia appointed a chief AI officer to lead upskilling initiatives.
• AI literacy and use‑case classification are central to responsible deployment.
• Teams must monitor agents’ access and daily activities to mitigate risk.
• Rapid agent development offers opportunities but demands strict oversight.

Summary

Georgia's state government treats AI as a collaborative responsibility, appointing a chief AI officer and embedding AI duties across security, legal, and technology teams.

The administration stresses AI literacy, classifying use cases as low, medium, or high risk, and trains staff to use only data essential for each scenario.

A key quote—"AI is not an individual sport; it's a team sport"—highlights the need for cross‑functional oversight while rapidly building AI agents and monitoring their access daily.

This model offers a blueprint for public‑sector AI governance, urging other jurisdictions to adopt rigorous oversight and monitoring to safely capture AI's benefits.

Original Description

At the NASCIO Midyear Conference, Georgia CIO Shawnzia Thomas described how the state is approaching AI leadership and literacy, making sure everyone has the tools they need and the skills to use them.

Video Transcript:

Well, in any state, AI is not an individual sport. It's a team sport. I named Nikhil Deshpande my chief AI officer a year and a half ago, but AI is not just his job. It's my job. It's my security officer's job, my legal team's job, my CTO's job. It's our job to make sure we understand the AI landscape, how we need to train, what tools we need to use. It's everybody's job to make sure they understand, you know, how it can help them to do their jobs easier.

I think AI literacy is the most critical characteristic of AI governance. Making sure that folks know how to use these tools, making sure they know which use cases to categorize as low, medium and high, which we do, and making sure they know what data to use. Only to use the data they need to use for that specific use case. So training them on AI, the tools makes responsible use much better. Because when they're able to use the tools responsibly, then we don't have to worry about issues coming out.

How are you exploring AI agents to supplement state IT work?

With these tools you can build agents quickly now. And so what I'm trying to get my team to do now is, let's make sure we understand what's being built, make sure we understand what level they're, they're being built to, make sure we understand what they have access to.

'Cause you can't give these agents access to everything. So wrapping our hands around that, getting a tool to make sure we can look at that on a daily basis throughout state government to make sure we know what these agents are doing. There's a lot of possibilities in using agents. There's also a lot of risk in using agents.