The biggest AI mistake has nothing to do with the technology

Most organisations are now developing AI governance frameworks. Far fewer have a framework for how AI affects the work itself and the people doing it.

There’s been real progress on the first part. The National AI Centre's Guidance for AI Adoption sets out six practices for responsible AI, including accountability, risk management, transparency, testing, and human oversight. These are undoubtedly a necessary part of any AI rollout.

But many organisations are still treating AI as another technology implementation. They're bolting AI onto existing jobs rather than redesigning the work itself. What happens to a role once half its tasks are automated? Will people trust the organisation enough to flag when AI gets it wrong? And is anyone genuinely getting time back to do more meaningful work, or are people simply absorbing the same workload with fewer hands?

Risk governance doesn't answer those questions - but change governance does.

The difference between AI transformation and AI whitewashing

A lot of what the media reports as AI transformation is simply cost reduction with a more palatable name attached.

Organisations claim productivity gains and efficiency wins, but a closer look often reveals headcount reductions with no meaningful redesign behind them. Without understanding work at the task level, many also struggle to demonstrate where AI has actually created productivity or delivered a measurable return on investment.

The organisations that get measurable value from AI often start at the task and job levels. They ask: What is actually involved in performing this role? Which tasks can be automated outright? Which can be augmented, where AI removes repetitive work while a person still exercises judgment? And most importantly, what higher-value work does that create capacity for?

As David Guazzarotto, Senior Principal at Mercer, put it at the recent Dayforce Summit Sydney: "This isn't about jobs disappearing. It's about the work itself being rewritten." Reinventing work is harder than eliminating roles, but it's also where new value is created.

Guazzarotto's view is borne out in Mercer's Global Talent Trends research. Sixty-three per cent of C-suite leaders globally say redesigning work for AI is the highest-ROI people investment available to them. Yet only one in three believes their organisation is equipped to do it.

A different approach to AI and redesigning work

One of the clearest examples of redesign done well is Ingka Group, the largest IKEA franchisee. In 2021, they deployed an AI chatbot called Billy to handle customer enquiries. According to the Ingka Group, between 2021 and 2023, Billy answered 47% of all customer questions, around 3.2 million queries. The chatbot handled the repetitive 47%, while people focused on the 53% that genuinely required human judgement.

The quick win at that point might have been to reduce headcount in the customer service team, but Ingka chose differently. They reskilled roughly 8,500 call centre employees into roles in interior design advisory, online retail sales, and complex problem-solving. The result was an estimated €1.3 billion in additional revenue at the end of FY22, generated by people freed up to do work that made better use of their skills and judgement.

That's the difference between governing the technology and governing the change. The decision about what to do with the capacity technology has created is what separates this story from another workforce-reduction headline.

Trust is a governance control

If redesign is one missing governance discipline, trust is the other. Mercer's research found that just 19% of organisations explicitly factor psychological and emotional impact into their digital implementation strategies, even though 68% of HR leaders believe they're already addressing it. Concern about AI-driven job loss has climbed from 28% in 2024 to 40% in 2026, and 62% of employees say their leaders underestimate AI's emotional impact on them.

This gap shows up in other ways, too. A 2025 Writer and Workplace Intelligence survey of 2,400 employees found that 31% admitted to actively sabotaging their company's AI strategy, including feeding poor-quality data into AI tools or deliberately underusing them.

Building trust requires more than communication. It means involving employees in identifying AI use cases, testing them, refining them, and helping shape how work changes. Organisations that treat AI as something to co-design, rather than something imposed from above, are more likely to create the psychological safety needed for people to experiment, challenge outcomes, and improve results.

That's why the trust questions in your engagement survey deserve more weight than they typically receive. Low scores are an early signal that people don't feel informed, don't feel they have enough autonomy, or don't believe they have any real influence over how their work is changing. By the time attrition appears, trust has often already eroded.

As workforce futurist Dom Price shared at the recent Dayforce Summit Sydney, leaders have two ways of approaching change: through fear or through play. "Fear gets you compliance, play gets you discretionary effort, and the ideas and innovation come from people who feel safe enough to experiment."

Efficiency isn’t the whole story

Most AI success metrics look similar across organisations: cost reduction, efficiency gains, and automation rates. They're easy to report to boards, but they miss what's actually happening to people's work week to week.

Employees notice when the repetitive parts of their job disappear, and they have more room to think, plan, and solve more complex problems. None of that appears on a dashboard built around cost cuts.

"We are humans first. Organisations are just a collection of humans," Price reminded leaders at the Dayforce Summit. It's an easy reminder to agree with, but a much harder one to build governance around, because most frameworks are designed to manage organisational risk rather than human capability.

Price also argued that AI should amplify good work rather than automate poor ways of working. If organisations simply layer AI over inefficient processes, technology only helps them do the wrong things faster. The bigger opportunity is to improve how work happens first, then use AI to scale what already works.

Seeing both sides of the AI transformation equation

Standard AI governance frameworks help organisations manage the ethical and safe use of AI. But they rarely provide guidance on how people adapt to it, trust it, or ultimately do better work as a result.

The organisations that will pull ahead over the next 12 months won't simply deploy more AI. They'll redesign the work itself. That means looking closely at job content, roles, augmentation, and capacity, rather than treating AI as another technology project to bolt onto the tech stack.

For HR leaders, that means having visibility into both sides of the equation: the technical and compliance risks, and what's actually happening to people's work, trust, and capability as AI reshapes the organisation. That's where the real opportunity sits.
 

Hear more from our experts

Get grounded view of how the workforce is likely to evolve as AI matures — and what leaders should prioritise now to prepare.
Explore more at Dayforce Presents