The Commonwealth Bank's recent backtrack on AI-driven job cuts tells a story that every Australian boardroom needs to hear. In July 2025, CBA announced 45 customer service roles would become redundant following the introduction of their AI chatbot, claiming the technology had reduced call volumes by 2,000 per week. By August, those same roles were reinstated. The AI hadn't delivered what was promised. Call volumes had actually increased, staff were working overtime and team leaders were being drafted to answer phones.
This wasn't a story about technology failure. This was a story about implementation failure.
As someone who has supported workplace change across France and Australia for more than two decades, I recognise the pattern. The CBA's experience isn't an outlier. It's emblematic of a broader challenge facing Australian executives as they navigate the promises and perils of artificial intelligence.
The Pattern Repeats Itself
CBA isn't alone in this experience. The Finance Sector Union's intervention forced the bank to admit 'we did not adequately consider all relevant business considerations' when announcing the redundancies. But look beyond Australia's shores and the pattern becomes clearer still.
Swedish fintech Klarna made headlines by replacing 700 customer service roles with AI, claiming their chatbot could handle the work equivalent of 800 employees. By mid-2025, CEO Sebastian Siemiatkowski was admitting to Bloomberg that the aggressive AI shift 'led to lower quality' and announcing rehiring initiatives. The company that once proclaimed 'AI can already do all of the jobs that we as humans can do' discovered that customer satisfaction had plummeted and that human interaction remained essential.
IBM laid off approximately 8,000 workers in 2023, replacing much of its HR division with AskHR, an AI-powered digital employee. The system performed routine queries and documentation adequately enough. What it couldn't do was handle tasks requiring empathy, subjective judgement, or nuanced human understanding. IBM has been quietly rehiring, with CEO Arvind Krishna acknowledging that despite extensive AI adoption, the company's total workforce has actually increased.
McDonald's spent three years developing AI-powered drive-thru ordering with IBM before ending the partnership in 2024. Viral videos of the system adding 260 Chicken McNuggets to orders while customers pleaded 'please stop' became the public face of an implementation that hadn't adequately accounted for real-world complexity. The technology worked in controlled environments. It failed in messy reality.
The Real Cost of Getting It Wrong
For Australian boards and CEOs, these cases reveal something more significant than technological limitations. They expose the cost of treating AI implementation as primarily a technology or efficiency question rather than a change management challenge.
Consider what CBA's approach cost them beyond the immediate embarrassment. There's the reputational damage, not just with customers but with the workforce that watched colleagues lose jobs based on flawed assumptions. There's the operational disruption of reversing course. There's the lost time and resources that could have been invested in more thoughtful implementation. And there's the erosion of trust that makes future transformation initiatives harder to execute.
The pattern across these failures reveals three consistent mistakes:
1 -> Treating implementation speed as a virtue in itself. The rush to demonstrate AI leadership drives premature decisions. Klarna's CEO pushed staff to rely on AI to fill gaps left by departing colleagues before the technology was ready. CBA made redundancy decisions before adequately assessing actual demand. McDonald's scaled a system that hadn't been sufficiently tested against edge cases. In each instance, the imperative to move quickly overwhelmed the discipline to move thoughtfully.
2 -> Underestimating the complexity of human work. The allure of AI often lies in its promise to automate the routine, freeing humans for 'higher-value' work. But this framing fundamentally misunderstands what makes human work valuable. Customer service isn't just information transfer. It's emotional labour, contextual understanding, relationship building. These aren't separate 'higher-value' functions layered on top of routine tasks. They're interwoven throughout the interaction. Strip away the human element and you don't just lose efficiency; you lose efficacy.
3 -> Failing to implement adequate measurement and feedback loops. CBA claimed call volumes were falling; union members reported the opposite. The gap suggests measurement systems weren't capturing actual operational reality or that feedback mechanisms weren't functioning. Effective implementation requires not just metrics but processes that allow ground truth to challenge executive assumptions.
What Australian Executives Must Do Now
Here's what separates successful AI implementation from expensive failure: discipline.
Start with the right question. Don't ask 'How can we use AI?' Ask 'What specific operational friction would AI meaningfully reduce?' This shift from technology-led to problem-led thinking prevents solutions searching for problems.
Implement rigorous pilots before scaling. Not showcase projects designed to confirm predetermined conclusions, but honest tests with clear success metrics, adequate timeframes and willingness to face uncomfortable truths. Measure not just efficiency but effectiveness. Track not just costs but quality outcomes.
Keep humans in the loop. The most successful implementations augment human capability rather than replace it. They handle routine tasks so people can focus on complex problems, but they maintain human oversight at critical decision points. IBM's AskHR experience proves why: the routine and the complex aren't separable in practice.
Treat AI as organisational change, not technology deployment. This means stakeholder engagement from the start. Clear communication about intent and impact. Bringing people along rather than imposing solutions on them. The Finance Sector Union wouldn't have needed to challenge CBA's redundancies if workers had been meaningfully involved in the assessment.
Build feedback mechanisms that challenge assumptions. CBA claimed call volumes were falling; union members reported the opposite. That gap reveals broken feedback loops. Create systems where ground truth can challenge executive assumptions before decisions become irreversible.
The Real Conversation
Australian executives face a choice. You can race competitors to implement AI quickly, or you can lead by implementing it properly. These aren't the same thing.
Real leadership means moving deliberately when market pressure demands speed. It means prioritising outcomes over timelines. It means having the courage to get it right rather than getting it done first.
The alternative? Join the growing list of organisations forced to backtrack, apologise and rebuild trust. Damage your reputation and operational effectiveness pursuing efficiencies that prove illusory. Treat employees as expendable based on optimistic assumptions that don't survive reality.
After supporting organisations through significant change across France and Australia for more than 20 years, I've learned that the hardest transformations aren't those involving new technology. They're those requiring us to change how we think about problems before we can solve them.
The conversation Australian boards need isn't 'How quickly can we implement AI?' It's 'What does responsible implementation look like?' Not 'How many roles can we replace?' but 'How can we improve outcomes for customers, employees and shareholders?'
These questions can't be answered by technology vendors alone. They require deep understanding of your organisation's culture, operations and people. They require honest assessment of risks alongside opportunities. They require the discipline to move deliberately despite pressure for speed.
CBA got it wrong, admitted it and corrected course. That takes courage. What takes more courage is learning from their example before making the same mistakes.
The technology will advance. AI will transform how we work. The question is whether we'll approach that transformation with appropriate seriousness, or mistake speed for progress and efficiency for effectiveness.
Australian executives can lead differently. Speed without wisdom isn't leadership. Efficiency without effectiveness isn't progress. Technology without humanity isn't the future any of us want.
You have the opportunity. The question is whether you'll take it.
TRICHORE TECH
What does responsible AI implementation actually look like? How do we navigate technological change without losing sight of the humans it's meant to serve? After witnessing major transformations touching Australian businesses, four Perth independents united by a shared vision gave birth to TRICORE TECH. Each bringing distinct expertise, they converged around fundamental questions: how to offer holistic approaches where technological change accounts for human dignity and organisational empathy, and how their combined experience could guide organisations through transformation responsibly.
TRICORE TECH helps Australian organisations analyse needs, diagnose situations, navigate organisational change and build, implement and maintain technology solutions under a unique ethical framework.
As signatories to the International Charter for Inclusive AI and committed to Australian AI Ethics Principles, we integrate AI and custom ERP systems with human-in-the-loop oversight, ensuring technology serves people rather than replacing them.
If you're navigating AI implementation or organisational transformation, get in touch: https://www.tricoretech.com.au/contactus
