Why Mainstream Thinking Could Kill AI Adoption In NZ Business (And Why The Younger Generations Might Save It)

AI’s ability to reason, adapt, and generate new solutions is both its greatest strength and its biggest challenge. Unlike traditional software, where the same input always gives the same result, AI agents are non-deterministic. They can give different answers to the same prompt, which can erode trust, especially in industries where consistency is critical.

Some functions will thrive with this flexibility (e.g. creatives), while others will struggle, particularly where teams are used to thinking in terms of “right vs wrong”.

The real opportunity lies in moving past this binary mindset. Curiosity and fresh thinking from AI-native generations will be key to unlocking AI’s true value.

Businesses are built for precision. Traditional systems are predictable: if something goes wrong, you can trace and fix it. AI is different. The same input can lead to different results depending on data, context, or how the system is set up.

This unpredictability is especially tough in sectors like supply chain, compliance, or healthcare, where consistency is vital. A shipping plan that changes every time, a compliance tool that flags different transactions, or a medical model giving conflicting results. All of these erode confidence.

When reliability is essential, it makes sense to design AI to behave more predictably. This can be done through better context, structured tools, or prompt design. In many cases, existing automation tools could handle these needs. AI is just pushing businesses to fix inefficiencies that should have been addressed long ago, and is able to deliver it at scale.

But to lead in AI adoption, companies need more than technical fixes. They need people who question assumptions, challenge outdated processes, and see non-deterministic outputs as opportunities, not failures. Younger AI-native thinkers are naturally suited to this.

Non-determinism itself is not new. Businesses have always dealt with unpredictable markets and shifting customer needs. What feels different now is that we’re so used to predictable software that AI’s behaviour feels foreign. If we focus only on making AI predictable, we risk missing its bigger potential.

The biggest barrier to AI adoption is not non-determinism itself but how leaders see it. Non-determinism is not a flaw. It is the point. It is what makes AI powerful, creative, and adaptable.

Now, I am not claiming to be in boardrooms with CIOs and CTOs debating this problem. I’m an enthusiast, speaking to those working with AI, following online conversations, and experimenting myself. I often wonder if we have been thinking about AI the wrong way. Should we, in fact, expect to know exactly what we want from AI?

Most AI adoption struggles come down to leadership and education: classic change management issues. If leaders don’t see the value in AI’s unpredictability, they’ll see it as a problem rather than an advantage. I wrote about this mindset shift in Agentic AI Is Not Failing, We're Just Judging It Wrong, where I argue that adaptability, not perfection, is what really matters.

Right now, too many companies are forcing AI into old-school workflows. IT teams treat agents like normal software deployments, focusing on permissions and integrations. Business teams expect outputs to look like static reports, identical every time.

This mindset is understandable. We’ve spent decades with software that only does exactly what it’s told. But that approach doesn’t work with AI. When we overload AI with rule-heavy prompts and “don’t do this” instructions, we don’t necessarily make it better; we make it less effective.

Trying to force AI into old patterns limits its potential. We expect precision when we should be designing for resilience. We try to script behaviours instead of shaping outcomes.

Chris Dixon at a16z calls this “skeuomorphic thinking”, applying new technology to old ways of thinking and legacy systems. Early websites looked like newspapers. Early smartphones tried to mimic physical keyboards. We’re doing the same with AI, asking “How can this upgrade what I already do?” instead of questioning whether those tasks are worth doing at all.

Generative AI challenges the whole “objective → workflow → output” model. When we give AI an objective but force it to follow our old logic, we’re holding it back. The real win comes from letting AI co-create, challenge assumptions, and open up entirely new ways of working.

AI works best when it has room to think. Its unpredictability is what allows it to reason, explore, and find answers we might not see ourselves.

The goal isn’t to control or tame AI but to learn how to work alongside it. Agentic systems thrive when we adopt a mindset that values curiosity over certainty, embraces iteration instead of chasing perfection, and prioritises feedback over finality.

Most companies aren’t built like this, which is why early AI projects often feel clunky. We give AI a goal but don’t give it the freedom to explore. We end up boxing it in, when its real strength is finding unexpected paths.

AI’s real value is in handling ambiguity, following intent, and producing creative results that were once too hard or expensive to achieve. We need to stop treating AI like a machine and start treating it like a collaborator.

This is why AI-native thinkers, people who aren’t burdened by decades of “this is how we do it”, are so important. They’re naturally more curious, experimental, and open to exploring new possibilities. Companies that embrace this will be the ones that find breakthroughs instead of just small improvements.

The idea of AI as a “digital teammate” isn’t new, but it’s still widely misunderstood. Too many companies treat AI like digital labour, expecting it to replicate existing human tasks with mechanical precision. We give it an objective, then over-engineer the instructions to achieve it, as if AI were just another bot running through a checklist.

A true teammate does not follow rigid scripts; they surprise you, challenge your assumptions, and bring fresh perspectives. AI should be treated the same way. If we keep forcing AI into old workflows, we will only achieve incremental improvements instead of the breakthroughs agentic systems can deliver. The digital teammate metaphor only works if we stop expecting AI to be deterministic. Like any teammate, AI learns through coaching, feedback, and iteration, improving with context, guidance, and examples of desired behaviour.

This shift is cultural as much as it is technological. Businesses need people, especially AI-native thinkers, who are comfortable guiding AI, refining outputs, and experimenting without needing to control every step.

The biggest cultural shift in AI adoption will come from moving away from deterministic, process-driven thinking. More experienced generations (Millennials included) are conditioned to work through the “4,000-hour apprenticeship", where expertise is built by grinding through repetitive tasks and learning through mistakes. AI is dismantling that pathway.

This raises real risks for young people. Entry-level roles, once the foundation of career growth, are disappearing. That’s not a small problem, and we can’t ignore it. But alongside this disruption lies an opportunity: younger, AI-native thinkers are uniquely positioned to lead this next era.

Those who have been in the workforce for decades might know how to apply AI to optimise existing processes, but the true potential of AI goes far beyond efficiency. It requires a mindset unencumbered by “this is how we’ve always done it". Younger people, who are growing up with AI as a natural extension of how they think and work, will be better equipped to challenge outdated workflows and imagine what’s possible.

This isn’t about token hiring. If enterprises want to lead in AI adoption, they need to bring these voices into the room early: as co-creators, not spectators. These are the people who can expose inefficiencies, experiment with new approaches, and push AI beyond simple automation.

It’s also part of a deeper shift, from traditional knowledge work to what Joe Hudson calls wisdom work. Knowledge work was about producing outputs: the perfect report, the polished spreadsheet. Wisdom work is about knowing which questions matter, when to act, and how to create the space where the best ideas can emerge. Younger generations, with their fresh perspective, can be the catalyst for this shift.

I don’t have all the answers: and that’s the point. We’re entering a phase where success with AI won’t come from perfect workflows or rigid plans. It will come from curiosity, experimentation, and a willingness to rethink the way we work.

Yes, speeding up back-office processes and automating repetitive tasks is essential. Many businesses will find that 90% of their “AI wins” at first are simply automating things they could have streamlined years ago with traditional software. That’s fine, it’s a necessary starting point. But it’s not where the real breakthroughs will come from.

The bigger opportunity lies in letting AI expose inefficiencies, challenge assumptions, and open up entirely new ways of thinking about value creation. This requires more than technical solutions. It requires human capabilities. People who are willing to ask different questions, experiment with new approaches, and guide AI like a collaborator rather than a tool.

This is where younger generations can be the superpower. Fresh graduates, AI-native thinkers, and those unburdened by legacy workflows are better placed to experiment without fear of “how things should be done". Their curiosity and adaptability give them an edge, and businesses that bring them into the room early will be far better positioned to uncover opportunities others miss.