Go Back
Franklin Williams, Chief AI Officer, Tyler Technologies | The CEO Magazine

Governing AI

In Focus
NAME:Franklin Williams
COMPANY:Tyler Technologies
POSITION:Chief AI Officer
For Tyler Technologies’ Chief AI Officer Franklin Williams, responsible AI adoption in government starts with strong safeguards, clear accountability and a commitment to trust from the beginning.
AI-generated summary

Nearly two-thirds of Americans believe AI is developing too quickly, according to an October 2026 poll from The Associated Press-NORC Center for Public Affairs Research. Meanwhile, around eight-in-10 believe that it’s ‘extremely’ or ‘very’ important for the United States government to keep AI under human control.

For Franklin Williams, Chief AI Officer at Tyler Technologies, one of the largest software and technology providers in the United States solely focused on the public sector, such findings highlight the core challenge facing government leaders: how to harness AI while maintaining public trust.

“Trustworthy AI in government starts by recognizing that trust cannot be added after deployment,” Williams tells The CEO Magazine.

From the outset, he explains, this means establishing clear governance, strong data protections, rigorous testing and transparency, alongside meaningful human oversight and accountability for AI outputs.

“Government agencies should also look for evidence that systems can handle regulatory nuance, refuse incorrect information and that they have been tested for bias and fairness,” he adds.

Starting small

Williams, who took up the role in May 2026 after serving as President of Tyler Technologies’ Data & Insights Division and Deputy CTO, says governments have a different set of responsibilities from private companies when deploying AI, because their relationships with residents are fundamentally different from those between businesses and their customers.

“Residents often cannot opt out of government services, and agencies hold highly sensitive information that people are required to provide,” he explains.

“Trustworthy AI in government starts by recognizing that trust cannot be added after deployment.”

That means governments should put governance, accountability and human oversight in place before deployment rather than treating them as safeguards to add later. When it comes to implementation, he advocates starting small — and with the problem, not the technology.

“Rather than immediately applying AI to the highest-risk government decisions, agencies can begin with high-impact, low-risk opportunities,” Williams says.

Answering repetitive questions, handling data entry, routing routine approvals or managing workflows are examples of uses that provide a practical environment to learn, establish governance and build institutional confidence before moving toward more complex uses.

Know your data

Even before AI touches anything, Williams recommends that governments build an inventory that identifies what data they hold, how sensitive it is, who owns it, where it lives and how long it is retained.

Historically, compiling such an inventory has been a lengthy process, although AI itself is starting to offer a solution.

“The first challenge is simply knowing what data the agency has,” he says. “We’re now seeing places where agentic AI can significantly shorten this process and quickly help governments build a very solid foundation for future AI development.”

“Governance should be strong enough to provide meaningful safeguards but flexible enough to evolve with the technology.”

Data sovereignty is particularly important in government as agencies need to retain control over their data and understand exactly where it goes and who can access it. That becomes even more important as AI agents take on more autonomous tasks.

“An agent moving sensitive resident information somewhere it does not belong is not simply a technical problem; it can become a breach of public trust,” Williams points out.

Progress, not perfection

One of the biggest misconceptions about responsible AI, Williams believes, is that governments must solve every governance question before moving forward.

“Similarly, the opposite extreme of moving fast simply for the sake of speed is not the answer either,” he says.

Instead, government leaders can establish governance based on existing best practices, start with low-risk, high-impact use cases, introduce checkpoints and adapt as they learn.

“Governance should be strong enough to provide meaningful safeguards but flexible enough to evolve with the technology and the agency’s experience,” he says.

“When trust is treated as the goal from the beginning, government can move forward with confidence while continuing to learn and adapt.”

Meanwhile, impact should be measured by whether workflows are improving, employees can spend more time on higher-value work and technology is delivering reliable results without compromising sensitive data or creating unacceptable risks.

After all, responsible AI is not about eliminating every risk or achieving perfection before taking the first step. Instead, it’s about building the systems, processes and accountability needed to manage risk while creating real value for employees and residents.

“When trust is treated as the goal from the beginning, government can move forward with confidence while continuing to learn and adapt,” Williams says.

 

Back to top