The majority of leaders are stuck in AI purgatory: Why pilots aren't making it to production

DXC Technology’s AdvisoryX says many AI projects are stuck on execution

Ninety four percent of business leaders face significant barriers when trying to deploy AI at scale, research from AdvisoryX, DXC Technology’s global advisory arm, has found.

The survey, Closing the AI Execution Gap, tapped 2,500 technology decision-makers across 22 countries. Kevin Jury, senior managing partner APAC at DXC Technology said 77 percent of leaders see AI as a board-level priority, but many organisations are still stuck in pilot mode for their AI programs, with far fewer making it into production in a meaningful way.

“One of the biggest challenges we see in Australia is disconnected experimentation. Teams launch pilots with good intent and the right technology in place, but without an appropriate operating model,” Jury said.

Because there’s no clear operating model, there’s also no clear enterprise roadmap for how these initiatives integrate with core systems, orchestrate across enterprise boundaries, evolve beyond a single use-case, or are governed over time, Jury noted.

Underestimating organisational change

Change management is also an issue with AI projects, with AdvisoryX finding a tendency for business leaders to underestimate the organisational change required. The survey revealed AI, in many organisations, continues to be viewed as a standalone initiative rather than a strategic enabler embedded across the business.

“AI can’t read between the lines and infer context, which can lead to technically correct but operationally wrong logic or inconsistent reasoning,” Jury said.

Organisations must understand AI will change some workflows, along with accountability, application design and decision-making, the survey found. These changes are particularly relevant in sectors where trust and oversight are critical, such as industries including financial services, healthcare and the public sector.

Jury added, “In these environments, shortcomings in data readiness, governance and workforce capability frequently becomes the blockers to scale.”

What business gets wrong about AI

Treating AI as a technology project, rather than an executive strategy, is an area where many organisations fall down. Some businesses also tend to treat AI as an efficiency play or point solution, meaning they’re automating what they already do.

Instead, organisations must consider and clearly define what success looks like, and how decisions are made, when AI and humans work together, Jury said.

“As a result, AI becomes a shortcut and not a multiplier that could provide richer context, deeper insights, better options, faster or situational awareness.”

Organisations also err in focusing their AI projects on cost compression, rather than expanding enterprise value. This value includes better customer experience, reduced losses, faster product cycles and enhanced capacity for innovation.

Many companies also confuse agency with automation, meaning they think AI is good for automating rigid tasks, but fail to realise AI agents can reason, manage exceptions and improve over time.

“Another misstep is assuming the tech teams alone should own AI, a misconception held by 73 percent of the leaders we surveyed.

“Executive strategy and cross-functional collaboration are critical to success, especially in areas such as compliance, ESG reporting, and R&D, where AI adoption is growing fastest,” Jury observed.

The AI opportunity

Jury said when end users are brought along on the journey, AI becomes a multiplier of capability and innovation that doesn’t simply automate but thinks, learns, adapts and performs at a fundamentally higher level.

And AI can help people make faster, better-informed decisions, instead of simply being an automation tool. This is particularly the case in areas like compliance, data-heavy operations and customer experience, the survey found.

“Our research shows that 81 percent of executives expect AI to increase workforce demand by 2028, especially in IT, data, cybersecurity, and software development,” Jury said.

“For Australian enterprises, it’s also a chance to address skills shortages by upskilling people for future-fit roles, lifting productivity and redesigning work around AI-human collaboration for more adaptive, resilient ways of working.”