AI or Die: Why Government Can’t Afford to Fall Behind

Original broadcast 6/17/25

Presented by EY

Artificial intelligence is no longer a futuristic aspiration—it's now an operational necessity in government. In the premiere episode of “Modernizing Government: The EY Insight,” filmed at the EY Center for Government Modernization in Washington, DC, Francis Rose explores how AI is driving transformation across federal agencies. Segment one features Amy Jones, EY’s U.S. Public Sector AI Lead, who unpacks the evolving risk-reward mindset shaping AI deployment. In segment two, Jones is joined by Charles Worthington, Chief Technology Officer and Chief AI Officer at the Department of Veterans Affairs, to dive into how VA is implementing generative AI tools like VA GPT at scale. Together, they outline not just the technology—but the culture shift—required to fully embed AI into government workflows.

Redefining Risk: Unlocking AI’s Value Across Government

Amy Jones, U.S. Public Sector AI Lead at EY, sets the stage for the conversation around the existential imperative facing government agencies: adopt artificial intelligence, or risk falling behind. But the “AI or Die” premise, she explains, isn’t about doomsday scenarios—it’s about modernization. Specifically, the stark difference between legacy systems that are no longer scalable and AI-powered approaches that unlock new capabilities.

Jones notes a fundamental shift in how agencies are evaluating AI. Historically, use cases were filtered first through a lens of risk, which narrowed the field significantly. Now, the funnel has flipped. Agencies are first identifying high-value opportunities and then assessing how to implement them responsibly. This reversal is unlocking a broader spectrum of use cases that align more directly with mission objectives.

She highlights the importance of moving beyond pilots and sandboxes, toward full-scale enterprise deployments. According to Jones, successful scaling requires a return to the basics—strong code architecture, governance, policy alignment, and budget support. Chatbots, for instance, should not be built in silos; instead, they should act as unified front ends to integrated back-end systems.

Crucially, AI is no longer confined to tech teams. The growing familiarity with tools like ChatGPT is bridging communication gaps between business and IT units. Non-technical users now better understand AI’s potential, enabling co-investment in transformation efforts. This shared language is laying the foundation for AI to become embedded in everyday government workflows, particularly as agencies face evolving public demands and constrained human resources.

Jones concludes by stressing that AI is not just a technological advancement—it’s a cultural one. Agencies must shift not just their systems, but also their thinking, to fully leverage the potential of AI.

Key Takeaways

• Agencies are now prioritizing value-first AI use cases, evaluating risk afterward.

• Scaling requires strong backend support and governance, not just flashy interfaces.

• Cross-functional collaboration is crucial to embedding AI into government operations.

Scaling with Purpose: Embedding AI into Government Workflows

In the second segment of “AI or Die,” Charles Worthington, Chief Technology Officer and Chief AI Officer at the Department of Veterans Affairs, joins Amy Jones to offer a real-world example of what AI at scale looks like in government. The conversation centers on VA GPT, the agency’s internal generative AI tool designed for secure use with VA data.

Worthington explains that VA GPT emerged in response to employee demand. As staff began experimenting with commercial tools like ChatGPT, VA recognized the opportunity—and the risk. In response, it built VA GPT on commercial models but within a secure, VA-approved environment. The result has been explosive: over 45,000 employees have already opted into the platform, saving an average of three hours per week and reporting greater job satisfaction. More importantly, the tool is helping employees eliminate tedious tasks and focus on work that requires their specialized skills.

VA GPT is far from the agency’s first foray into AI. The VA has a robust AI inventory with over 200 active use cases, ranging from clinical diagnostics and imaging to cybersecurity and administrative workflows. Worthington describes how the success of VA GPT builds on this foundation but distinguishes itself through scale and flexibility.

Amy Jones praises the VA’s approach, particularly its emphasis on empowering frontline employees. By enabling decentralized experimentation, the VA is surfacing use cases from the people who know the pain points best. This grassroots innovation is supported by a growing community of practice where staff share prompts, create micro-apps, and build reusable tools—accelerating adoption across the enterprise.

Jones also highlights the value of “skill stores,” internal repositories of best practices and templates that standardize effective workflows. These stores reduce the need for repeated software development and help ensure consistent, secure implementations.

Worthington adds that scaling slowly and thoughtfully has been key to the program’s success. Rather than pushing full access too quickly, the VA prioritized iterative rollouts, learning from user behavior along the way. Their goal is to expand VA GPT access to the entire workforce by year’s end.

The conversation then turns to the future. Worthington believes that today's chat-based AI interfaces will eventually look antiquated. He envisions AI deeply embedded into core workflows, from scheduling and claims processing to software development. In fact, VA is already using AI to enhance systems like appointment optimization, which has helped over 300,000 veterans get earlier care.

From a strategic perspective, both Worthington and Jones emphasize the importance of marrying technology with business insight. Worthington points out that successful projects start with clearly defined KPIs—measurable improvements like processing time or call volume—not extensive feature lists. And while many agencies still digitize analog processes, Jones argues the next leap is in rethinking those processes entirely, leveraging AI’s full automation potential.

One striking example is software development. Worthington, a software engineer himself, shares how AI has revolutionized coding. Developers can now prompt generative tools to write, test, and deploy code across multiple systems. This capability could be transformative for agencies saddled with legacy systems written in outdated languages.

Finally, both guests stress the importance of strong communication between business and technology teams. Misalignment often occurs not due to malice, but due to poor translation between roles. Rapid development cycles and constant feedback loops are helping overcome this, creating tighter, more agile partnerships.

In the end, Worthington and Jones agree: AI is about more than saving time—it’s about freeing up the cognitive space necessary for transformative thinking. Whether that’s reimagining scheduling systems, modernizing software delivery, or redefining claims processing, the opportunity is vast—and the time to act is now.

Key Takeaways

• VA GPT empowers frontline employees with secure, generative AI—saving time and improving satisfaction.

• Scaling is driven by grassroots innovation, supported by communities of practice and reusable micro-apps.

• The future of AI in government lies in rethinking workflows, not just digitizing them.

This is the first in a new limited series: Modernizing Government: The EY Insight. As society and technology continue to advance rapidly, the imperative for government modernization is more pressing than ever. "Modernizing Government: The EY Insight," hosted by Francis Rose, offers a comprehensive look into how governments can evolve to become more efficient, modernized, and responsive. Set within the EY Center for Government Modernization, this special brings together a compelling mix of government leaders, EY executives, and strategic partners for an in-depth discussion on the road to transformation.