AI Strategy vs AI Reality: What Irish and UK Leaders Are Getting Wrong

The Gartner hype cycle has placed generative AI in the "trough of disillusionment." After the initial rush of excitement, many organisations are discovering that AI pilots fail at alarming rates. Reports suggest that 95% of AI initiatives do not deliver the promised results. In this episode of Chatting GPT, host Mary Rose Lyons speaks with Stephen Redmond, Head of Data and Analytics and AI at BearingPoint and former Accenture AI leader, about why AI strategy is failing and what leaders can do differently.

Why AI Pilots Fail

Redmond is clear about the primary cause of AI pilot failures: poor goal-setting. "Most AI pilot failures stem from poor goal-setting, not technology limitations," he explains. Organisations are deploying AI without clearly defining what value they expect to create.

This problem is compounded by the hype cycle itself. When AI was at the peak of inflated expectations, organisations rushed to implement solutions without understanding their own needs. Now, as reality sets in, many are discovering that their pilots were solving the wrong problems—or no real problem at all.

The solution, Redmond argues, is to define value before deployment. Organisations must understand their business challenges deeply before selecting AI tools to address them. This requires patience and discipline that the hype cycle often discourages.

The Shadow AI Problem

One of the most significant challenges Redmond identifies is shadow AI—the use of AI tools by employees without organisational approval or oversight. Even when companies lock down corporate systems, employees are using ChatGPT on personal devices to get their work done.

This creates serious data security risks. Sensitive company information is being processed through consumer AI tools with unknown data handling practices. Compliance frameworks are being bypassed. And organisations have no visibility into what AI tools their employees are using or how.

Redmond's insight is that shadow AI is not primarily a technology problem—it is a literacy problem. Employees turn to unauthorised tools because they do not have access to approved alternatives and do not understand the risks. Addressing shadow AI requires giving employees better options and educating them about safe usage.

Literacy Before Strategy

Perhaps Redmond's most important insight is that AI literacy must precede AI strategy. When staff understand AI tools safely and creatively, better use cases emerge from the ground up rather than being imposed from the top down.

Many organisations attempt to define AI strategy through executive workshops and consultant-led planning sessions. Redmond suggests this approach often misses the mark because it lacks input from the people who will actually use the tools. Frontline staff often have the best understanding of where AI could create value—but only if they understand what AI can do.

"When staff understand tools safely and creatively, better use cases bubble up from the ground," Redmond notes. This bottom-up innovation is essential for finding the high-impact applications that top-down strategy often misses.

The Show-Tell-Ideate Approach

Redmond recommends a practical three-step approach for building AI capability: show, tell, ideate.

First, demonstrate AI tools in action. Let employees see what is possible with current technology. This builds intuition and sparks creative thinking about applications in their own work.

Second, share industry examples. Show how similar organisations are using AI successfully. This provides context and helps employees understand what good AI implementation looks like.

Third, run structured ideation sessions. With employees now informed about AI capabilities, facilitate workshops to identify use cases specific to your organisation. This produces prioritised, high-return opportunities tied to real business needs.

This approach reverses the typical strategy sequence. Rather than defining strategy first and then implementing, it builds capability and understanding first—allowing strategy to emerge from informed practice.

Moving Forward

For organisations in Ireland and the UK, Redmond's message is both cautionary and hopeful. The trough of disillusionment is real, but it is also an opportunity. Organisations that take the time to build literacy, define value clearly, and involve their people in the process will emerge stronger.

The alternative—continuing to chase hype without building foundation—is a recipe for continued failure. As AI continues to evolve, the gap between organisations that invested in literacy and those that did not will only widen.

Want the full conversation? Watch the Chatting GPT episode on YouTube here: https://www.youtube.com/watch?v=cdL-f2WiBmc