September 28, 2026
Federal agencies do not need to completely modernize their technology environments before taking advantage of artificial intelligence, according to former FBI Chief Information Officer Gordon Bitko.
Bitko, now executive vice president for public sector at the Information Technology Industry Council, says some of the biggest barriers to AI adoption have less to do with the technology itself and more to do with legal requirements, organizational culture, business processes and acquisition.
“The biggest challenge frequently is not actually about the technology itself,” Bitko says.
Agencies have historically struggled to adopt emerging technologies, he explains, and AI presents many of the same challenges. But AI could also become part of the solution.
One example is data management. Agencies often have information distributed across numerous legacy systems, making it difficult to understand what data exists, who owns it and how it is being used. Bitko says AI can potentially help agencies map those assets, better understand their data environments and prioritize modernization efforts.
Rather than approaching AI through traditional technology deployment models, Bitko encourages agencies to identify specific pilot opportunities where they can safely test the technology and learn from existing federal use cases.
Acquisition remains another challenge. Bitko points to GSA’s OneGov approach as an innovative way for the federal government to leverage its purchasing power, but individual agencies still need to determine how to use those tools effectively.
That requires bringing together the right people across procurement, cybersecurity and data. Each group needs to understand both the technology’s potential and the agency’s requirements.
AI adoption can also provide an opportunity to rethink existing business processes rather than simply applying new technology to old ways of working.
Bitko says successful modernization efforts often begin with finding a mission partner willing to experiment on a small scale. Agencies can learn from that experience, demonstrate results and then expand successful approaches across the enterprise.
Scaling introduces its own challenges.
Early adopters may be comfortable experimenting with technology that does not always work perfectly, but employees across an entire organization generally expect tools to work immediately. That makes workforce training and application resilience increasingly important as agencies move from pilots to enterprise deployments.
Success can create another challenge: demand. Bitko recalls that as cloud pilots expanded, greater-than-expected usage created new cost and resource requirements. Agencies need to account for similar possibilities as they scale AI.
Most importantly, Bitko says agencies should not wait until their infrastructure is completely modernized before moving forward.
Instead, they can identify use cases that take advantage of current AI capabilities while learning how to integrate those tools with legacy environments. In some cases, Bitko says AI itself could serve as an intermediary between modern capabilities and legacy applications or data.
As agencies measure their progress, Bitko points back to data. Understanding what data an agency has, where it comes from and who should have access provides a foundation for evaluating whether AI tools are improving over time.
“The models, the systems, the AI tools are only as good as all that data,” Bitko says.