Why Most Irish Businesses Are Still Playing It Safe with AI Innovation

A stark warning emerges from the latest research: only 1% of organisations globally feel their AI initiatives have reached maturity. For Irish businesses, this statistic isn't just a number—it's a mirror reflecting a widespread hesitation that could determine which companies thrive and which fade into obscurity like Blackberry or Xtravision.

In a revealing conversation on the Chatting GPT podcast, Lee Bristow, a Cape Town native now based in Ireland, unpacked why so many Irish businesses are 'dillydallying in the shallow end' when it comes to AI adoption—and what this reluctance could cost them in an increasingly competitive global marketplace.

The Three Categories of AI Adoption

Bristow identifies three distinct approaches Irish businesses are taking towards AI implementation, each with dramatically different implications for future competitiveness.

The first category comprises organisations paddling in the shallow end—those who 'lift and shift' minimal AI capabilities into isolated parts of their business. These companies often hide behind regulatory concerns, citing frameworks like the EU AI Act as barriers to innovation rather than guardrails for responsible implementation. According to Bristow, this approach represents the most dangerous position: appearing to engage with AI whilst fundamentally avoiding the transformative work required.

The second category takes a strategic lens, identifying specific high-impact areas based on business strategy. These organisations may have limited budgets, but they're willing to 'go deep' when they do invest, fundamentally retooling processes and capabilities to achieve their objectives through AI. This measured but committed approach demonstrates genuine strategic thinking.

The third category—represented by companies like Box, the US file storage system—are going 'all in' as AI-first businesses. These organisations are rethinking and retooling everything from the ground up, embedding AI into their core operating model rather than treating it as an add-on.

Bristow's assessment is unequivocal: 'Anyone that is in the first category, I think they're going to go the route of Blackberry.' The technology evolution won't wait for hesitant adopters, and once AI hits the top of the adoption curve, businesses without operational capabilities in this new paradigm simply won't survive.

Leadership and Culture: The Determining Factors

What determines which category a company falls into? According to both Bristow and podcast host Mary Rose Lyons, founder of AI Institute (Ireland & UK), the answer lies squarely with leadership and culture—two elements that are, in practice, inseparable.

'There's not necessarily a distinction between leadership and culture,' Bristow explains. 'The leadership becomes the culture of the organisation.' This observation carries profound implications for AI adoption. Leaders who remain cautious or dismissive about AI inevitably create organisational cultures that mirror that hesitation, regardless of what formal policies might state.

The challenge intensifies when leadership's reluctance collides with employees' practical needs. This collision creates what Bristow calls 'shadow AI'—the unauthorised use of AI tools by employees who need them to perform their jobs effectively, even when official policy prohibits such use.

The Shadow AI Crisis: 78% and Growing

Perhaps the most alarming insight from the conversation concerns shadow AI prevalence. Reports suggest up to 78% of employees are using AI tools regardless of whether organisational policy permits it. For Irish businesses, this creates a paradoxical situation: by attempting to avoid AI risks through prohibition, leadership inadvertently creates far greater risks through ungoverned usage.

The GDPR implications alone are staggering. Employees uploading CRM databases to free versions of ChatGPT for analysis, sharing client information with unauthorised tools, or processing sensitive data through unsecured platforms—all of these represent serious data exfiltration risks that prohibition policies cannot prevent.

Bristow shared a compelling example from Volkswagen, where an engineer used AI to explore drivetrain designs. The system returned information about a VW model that was never supposed to reach market—a trade secret inadvertently revealed through an AI interaction. Whilst the specifics require verification, the scenario illustrates perfectly how intellectual property can leak into model training data when proper controls aren't in place.

The Enterprise Ireland Paradox

The conversation took a particularly pointed turn when discussing Enterprise Ireland, the government agency supporting Irish businesses. Despite national rhetoric about embracing AI—including statements from Pascal Donohoe in January 2024 supporting AI across the public sector—local enterprise offices remain banned from using any AI tools in their operations.

As Lyons observed, 'Here we are dealing with the part of government that engages with entrepreneurial minds and beings. What hope do we have for things like education and healthcare if that's the mentality?'

This disconnect between policy rhetoric and operational reality exemplifies the 'shallow end' approach. By prohibiting AI use without providing approved alternatives, Enterprise Ireland potentially exposes itself to the very shadow AI risks that proper governance would prevent. Bristow estimates that even in restricted environments, perhaps half of employees are using AI 'illegally'—without oversight, security, or governance.

The Strategic Solution: Approved Tools and Proper Governance

The solution, according to Bristow, isn't prohibition but strategic provision of appropriate tools. For organisations like Enterprise Ireland, this could mean leveraging existing Microsoft relationships to deploy Copilot—a solution with strict data handling rules that prevents information from being used in model training.

'All you have to do is provide the required tools,' Bristow argues. The investment in approved, secure AI capabilities is infinitely smaller than the potential costs of data breaches, intellectual property loss, or GDPR violations resulting from shadow AI usage.

This approach applies equally to private sector organisations across Ireland. Leadership must move beyond viewing AI as a risk to be avoided and instead recognise it as a capability to be governed. This means establishing clear policies, providing approved tools, implementing proper controls, and ensuring the vision for AI adoption is embedded throughout the organisation through effective governance frameworks.

Moving from the Shallow End to Strategic Depth

For Irish businesses still hesitating, the message is clear: the competitive landscape is shifting rapidly, and shallow engagement with AI represents the highest risk position of all. Organisations need not immediately become 'AI-first' businesses, but they must at minimum adopt the strategic approach—identifying key areas for deep implementation and committing the resources necessary to genuinely transform capabilities in those domains.

The alternative—continuing to paddle in the shallow end whilst competitors dive deep—leads to only one outcome: becoming another cautionary tale of disruption, joining Blackberry and Xtravision in the archives of businesses that failed to adapt to technological transformation.

The question for Irish business leaders isn't whether to engage with AI, but how quickly they can move from prohibition and hesitation to strategic governance and genuine adoption. The 1% who feel their AI initiatives have reached maturity didn't get there by playing it safe—they got there by going deep.

Want the full conversation? Watch the Chatting GPT episode on YouTube here: https://www.youtube.com/watch?v=HfbYMsV2vKQ