Taka Ariga says government must reimagine AI, not automate old systems

By AI, Created 4:06 PM UTC, May 29, 2026, /AGP/ – On the CAIO Connect Podcast, former GAO and OPM AI leader Taka Ariga argued that public agencies need to redesign workflows around AI, not layer new tools onto outdated processes. He said the real challenge is governance, workforce change and public trust, not just technical adoption.

Why it matters: - Public agencies are under pressure to adopt AI without weakening privacy, security or trust. - Ariga said government cannot copy Silicon Valley’s “move fast and break things” approach because failures can expose sensitive data and damage confidence in institutions. - The bigger opportunity is redesigning government operations for AI, not using AI to preserve old workflows.

What happened: - On the latest CAIO Connect Podcast episode, host Sanjay Puri spoke with Taka Ariga, founder of Sol Imagination LLC. - Ariga discussed how government should adopt AI responsibly while balancing innovation, governance, workforce concerns and public trust. - Ariga previously led AI and data initiatives at the Government Accountability Office and the Office of Personnel Management. - He argued that agencies must rethink their operating models instead of simply automating outdated processes.

The details: - Ariga described the Government Accountability Office as Congress’s watchdog agency, with audit work spanning healthcare, defense and education. - He called GAO a “candy store” for data scientists because of the volume of public data and policy work involved. - Ariga said the Office of Personnel Management manages more than 2.3 million civilian employees as the federal government’s human resources agency. - He said both roles showed him that AI in government is a policy, ethics and workforce challenge as much as a technology challenge. - Ariga said public-sector AI must account for privacy, security, data sovereignty and workforce anxiety. - He described launching the GAO Innovation Lab as building a startup-like effort inside a century-old audit institution. - The lab had to build cloud infrastructure, create secure AI testing environments and develop new procurement rules for emerging technologies. - Ariga said the hardest part was change management inside a skeptical organization. - “Everyone wants innovation, but nobody wants change,” Ariga said. - Ariga said middle managers are critical to AI adoption because they control workflows and influence employees directly. - He said government agencies should hire mission-driven talent even when public salaries cannot match major technology companies. - At GAO, Ariga created a flatter structure that let employees lead projects and work across technical and policy disciplines. - Ariga warned that many organizations focus on small productivity gains instead of end-to-end transformation. - He said agencies need to orchestrate multiple AI systems responsibly while keeping humans involved in critical decisions. - Ariga noted that workers must adapt to probabilistic AI systems whose outputs can change from one day to the next. - “The spreadsheet that you open on Monday is the same spreadsheet you open on Thursday,” Ariga said. “But with AI, the answer may be slightly different every time.” - Ariga said governance should accelerate innovation rather than block it. - He criticized slow approval systems and called for more agile oversight models that are updated as technologies evolve. - He said many companies invest heavily in AI without solving meaningful problems. - In Ariga’s view, successful AI adoption starts with the right use case, scalable systems and redesigned workflows for a digital-first future.

Between the lines: - Ariga’s comments point to a broader shift in public-sector AI thinking: the main bottleneck is no longer access to tools, but organizational readiness. - His emphasis on middle managers suggests AI adoption may succeed or fail less on strategy decks and more on day-to-day workflow control. - The podcast framed governance as an enabler of AI progress, not just a compliance layer.

What’s next: - Agencies are likely to keep testing AI use cases, but Ariga said the work must move from pilots to systems that can scale. - The next phase will require updated policies, better oversight and redesign of legacy processes around digital-first operations. - Ariga urged organizations to focus on practical impact rather than hype as AI adoption expands.