Innovating at AI Speed While Staying in Control: CIOs Face New Governance Demands – PwC

Enterprises are racing to capture the benefits of artificial intelligence, but speed comes with risk. Business units now have the power to generate code and digital assets on their own, raising urgent questions about how to maintain control without slowing innovation. That balancing act is redefining the role of CIOs, according to Dan Priest, US Chief AI Officer at PricewaterhouseCoopers (PwC).

Speaking during a BizTechReports executive vidcast and reflecting on discussions at this year’s CIO 100 conference in Scottsdale Arizona, Priest described how the push to innovate at AI speed is colliding with the need for governance, resource management, and strategic alignment. 

“The most successful technology deployments I’ve ever seen were those with strong business ownership,” he said. “But you don’t want code entering production in uncontrolled ways. CIOs have to both empower the business and govern the path to production more tightly than ever.”

The Speed vs. Control Equation

Generative AI has upended long-standing boundaries between IT and business units. Teams that once delivered requirements to IT can now bring AI-generated code alongside those requirements. This accelerates experimentation and solution-building but creates new risks around compliance, integration, and security.

“The business depends on technology, data, and AI to be competitive,” Priest said. “Naturally, stakeholders want more control. That’s healthy. But it means the value proposition of IT teams continues to evolve.”

For CIOs, the equation is shifting: less about centralized delivery and more about orchestrating safe, rapid innovation. “Scale matters less,” Priest said. “Speed matters more, and innovation matters most of all.”

Governance That Enables, Not Blocks

Governance sits at the heart of the tension. In the early cloud era, enterprises struggled with “shadow IT” as SaaS subscriptions appeared on expense reports without central oversight. Now, generative AI risks repeating that pattern.

“You no longer have to go through years of training to code,” Priest explained. “You can generate code using natural language prompts. That’s powerful, but you can’t have it entering production without controls. The path to production needs to be governed more now than ever.”

Traditionally, governance has been viewed as a funnel, narrowing the flow of projects and slowing time to deployment. In an AI-driven environment, Priest said, that perception has to change. CIOs must redesign governance to operate as a pipeline that accelerates safe innovation, not as a roadblock.

“Every function has the potential to disrupt itself,” he said. “CIOs have to govern differently — at the speed of business — while still ensuring assets are integrated, secure, and compliant. That balance is hard, but it’s essential.”

Pressure on Resources and Talent

AI adoption has also reshaped resource allocation. While AI-generated code may reduce development bottlenecks, it can create new costs in infrastructure, energy consumption, and oversight.

“Senior developers are some of the best at reviewing AI-generated code,” Priest noted. “But they don’t want to spend their day doing code reviews. They want to build interesting and valuable solutions. Roles need to evolve so developers are fulfilled while quality and compliance remain intact.”

This evolution will influence how CIOs allocate talent. Rather than treating AI-generated code as a finished product, enterprises must invest in validation and integration. Quality assurance becomes less about defect detection and more about ensuring AI outputs align with enterprise architecture and data flows.

Even iteration cycles are being redefined. Where CIOs once sought to minimize the number of revisions between requirement and production, they now accept, and even encourage, more iterations. “Multiple turns of the crank are fine if they lead to better quality code faster,” Priest said. “The efficiency of AI makes that possible. As a result, the metric shifts from how many cycles are completed to how much value is created.”

Realigning IT-Business Relationships

As AI accelerates the pace of innovation, the relationship between IT and business leaders is being reshaped. Profit-and-loss owners are increasingly generating their own digital assets, closer to the markets they serve.

“That’s not a bad thing,” Priest said. “The best code is often developed closest to where value is created. But change is hard, and CIOs must redefine where they add value.”

Not everything will change, however. According to Priest, certain responsibilities, like ensuring data quality, integrating code into enterprise systems, maintaining security, and scaling solutions across the enterprise, should remain firmly within IT. 

That said, IT’s role will expand to include translation, coaching, and governance. “The relationship is evolving quickly,” Priest added. “CIOs need to act less like service providers and more like strategic partners who can empower business units while keeping innovation safe and sustainable.”

Measuring Value at AI Speed

Priest emphasized that the speed vs. control equation ultimately comes down to value realization. “Are we saving money or making money? That’s the bottom line,” he said.

CIOs need metrics that reflect both innovation and governance. As an example, technology leaders should absolutely see the cost of code production decline as AI accelerates development. Similarly, organizations should carefully track adoption of AI tools across business units. And quality must be tracked through both traditional measures, such as defect rates, and new ones unique to AI.

Beyond hallucinations, AI agents sometimes produce “deceptions.” They may claim to have completed tasks when they did not actually execute. Monitoring and reducing such instances is critical. “That is why instrumentation is going to be so important. It creates a common language between IT and the business,” Priest explained. “It ensures governance is accountable to value realization, not just control.”

This new approach to metrics also redefines expectations. In the past, speed was measured by how quickly requirements moved through the pipeline. Going forward, efficiency will be measured by how much value each cycle delivers, even if the number of cycles increases. “The business wants results, not just throughput,” Priest said.

Beyond Hype to Nuance

Priest contrasted today’s discussions with those at the CIO-100 symposium a year ago. “Last year, there was a lot of breathlessness, even hysteria,” he recalled. “This year, the conversation is more nuanced. We’re talking about governance, resources, strategy, and value.”

That nuance, he argued, is essential for CIOs navigating the speed vs. control spectrum. It reflects a recognition that innovation cannot come at the expense of safety and compliance, but that governance cannot be allowed to stifle experimentation.

“The world will be powered by digital, and much of it will be created with AI,” Priest said. “Humans will lead, but AI will enable. CIOs should be excited about this era because it puts them at the center of value creation.”

###