AI adoption resistance: why your team won't use AI tools
TL;DR
The AI adoption narrative has been wrong from the start. Employees are not refusing AI because they're lazy or technophobic, they're refusing because the tools don't work, the mandates are toxic, and the change management has been almost universally neglected. A third of employees skip AI tools entirely. Duolingo's CEO publicly reversed his own AI mandate. Successful AI implementation is 80% people and 20% technology, and most organisations have that ratio completely backwards.
Is the Duolingo AI mandate a cautionary tale for every business?
Yes, and it's one every business leader should read carefully. Duolingo CEO Luis von Ahn went public a year ago with an "AI-first" declaration, announcing that employees would be evaluated on their AI usage during performance reviews. The message was unambiguous: use AI or your career here will suffer.
He had to publicly walk it back. "I'm not going to force you, " von Ahn conceded, because employees pushed back hard. They asked pointed questions about whether the company simply wanted them to use AI for the sake of using it, regardless of whether it helped them do their jobs better. They refused to have their professional competence measured by how often they opened a tool rather than by the quality of their actual output.
Von Ahn even admitted that AI-written code "can be difficult to debug and is not consistently reliable." The CEO of a prominent tech company acknowledging that the tools he was mandating were actively creating problems, not solving them.
Forcing employees to use AI through performance metrics or top-down mandates is a recipe for resentment, resistance, and ultimately failure.
Duolingo is not an outlier. A SAP and WalkMe survey found that a full third of employees skip using AI on tasks because it either disrupts their workflow or costs them more time than doing the work manually. One in three. These are professionals making a rational assessment that the tool is not helping them.
What actually happens when you mandate AI adoption?
You get performative compliance. People tick the adoption box to keep their managers happy while quietly doing the real work the old way. The promised productivity gains evaporate, morale crumbles, and you've spent serious money on tools that gather dust.
When you tie an employee's livelihood to their adoption of a technology they don't trust or find useful, the focus shifts from doing good work to gaming a metric. That's not a technology problem, it's a leadership problem. Worse, you create a culture where people pretend to use AI to satisfy reporting requirements while the actual business runs on the same old processes it always did.
Why are Australian frontline workers struggling with AI?
They're losing 13 hours every month just dealing with device downtime, connectivity issues, and manual workarounds, that's nearly two full working days a month wasted before AI even enters the picture.
A SOTI report on Australian frontline workers in transport, logistics, healthcare, and emergency services puts hard numbers on the problem:
- 64% of emergency service workers say technical issues add stress to their already demanding jobs
- 58% of organisations in distributed workforce environments still rely on manual processes like email and paper for critical operations
- Only 34% of these organisations have increased spending on mobile security, even as they pile more AI tools onto fragile foundations
Layering advanced AI tools on top of unstable, legacy infrastructure actively harms frontline workers and reduces productivity rather than improving it.
You cannot expect a delivery driver, a nurse, or a paramedic to embrace AI when the basic tablet or handheld device they use to access it barely functions. They don't see a revolutionary new tool; they see another thing that doesn't work properly that they're now expected to use on top of everything else. They are building a skyscraper on quicksand and wondering why it keeps sinking.
See where AI fits in your business. Free.
A 45-minute audit. We map the highest-value automations and what they're worth in time and money. No pitch, no pressure.
Is the AI skills gap really a talent pipeline problem?
No. The data is clear: it's a management failure, not a talent problem.
A six-country study by Pearson and AWS, covering the US, UK, Brazil, Saudi Arabia, Vietnam, and Malaysia, found:
- 53% of employers say their primary challenge is finding graduates with the right AI skills
- 78% of higher education leaders believe they are already meeting employer expectations
- Only 14% of graduates report high proficiency in applying AI tools professionally
Everyone thinks someone else is solving the problem, and nobody actually is. Universities think they're preparing students adequately. Employers think universities are failing them. Graduates enter workplaces where they're expected to use tools they were never properly trained on, in environments that were never set up to support them.
Pearson identifies what they call the "AI Readiness Friction Framework", six compounding frictions that prevent organisations from becoming truly AI-ready: pace, connection, capability, governance, experience, and skills. Notice that skills is just one of six factors. The other five are all management and organisational issues that have nothing to do with whether your team can operate a chatbot.
Who in the organisation actually needs the most AI support?
The people closest to the actual work, and they are almost certainly not getting it.
Grant Thornton's survey of 950 C-suite leaders found that frontline employees (37%) and middle managers (30%) are the people who need the most support to implement AI effectively. These are the people who will determine whether your AI strategy succeeds or fails in practice. Yet in most organisations, training budgets and change management attention flow upward to senior leadership, not downward to the people who matter most.
We are expecting people to inherently understand how to use complex new technologies without providing the necessary support, education, or stable environments to do so. The skills gap is not a pipeline problem from universities; it is a failure of leadership to manage the transition within their own organisations.
What does effective AI change management actually look like?
Successful AI implementation is roughly 80% people and 20% technology. The businesses getting real results are the ones that invest as much in change management as they do in the technology itself.
In practice, that means:
- Start with problems, not tools. Ask your frontline workers what slows them down, what frustrates them, what they wish they could do faster, then find AI solutions that address those specific pain points.
- Fix the foundation first. Broken devices, unreliable connectivity, and manual workarounds need to be resolved before you even think about introducing another layer of technology.
- Build practical training programs. Not a one-hour webinar followed by a login and a pat on the back. Hands-on, ongoing, with safe spaces to experiment without fear of judgement or consequences.
- Celebrate real solutions. Recognise the people who find creative ways to use AI to solve actual problems, rather than punishing those who haven't hit some arbitrary adoption metric.
- Be honest about limitations. Von Ahn's admission that AI-written code isn't consistently reliable is refreshing precisely because it's rare. Most leaders are still pretending AI is a magic wand, and their teams know it isn't. That credibility gap destroys trust and makes adoption even harder.
The resistance your team is showing is not irrational, it is feedback. A third of employees skip AI because it genuinely disrupts their workflow or costs them more time than doing the work manually. Listen to that signal instead of trying to mandate your way past it.
What to do this week
- Audit your AI mandate. Are you measuring tool usage or business outcomes? If you're tracking logins and adoption rates rather than output quality and efficiency gains, you're measuring the wrong thing. Fix the metric first.
- Talk to your frontline. Ask three people closest to the actual work what slows them down most. Don't pitch AI, listen. If the answer is broken devices, bad connectivity, or manual workarounds, fix those before you add anything new.
- Check where your training budget flows. Is AI training investment going to senior leaders or to the frontline employees and middle managers who need it most? Redirect it downward.
- Drop one AI mandate. If you have a policy tying performance reviews to AI tool usage, remove it this week. Replace it with an outcome-based measure and observe what changes.
- Map your friction points. Using Pearson's framework, pace, connection, capability, governance, experience, skills, identify which of the six frictions is the biggest blocker in your organisation right now, and address that one first before touching anything else.
Where to from here
Book a free 60-minute AI audit, we'll explore exactly what workflows are worth augmenting with AI.
Live with passion & AI,
Brett
Host a podcast? Have Brett on as a guest.
Straight talk on implementing AI in real SMEs, no jargon, plenty of receipts from the businesses we run.
Frequently asked questions
Why do employees resist using AI tools at work?
+
A SAP and WalkMe survey found that a third of employees skip AI tools because they disrupt their workflow or cost more time than doing the work manually. It's a rational response to tools that don't fit the actual job, not technophobia.
What happened when Duolingo mandated AI use?
+
Duolingo CEO Luis von Ahn publicly backtracked on his AI-first mandate after employees pushed back against being evaluated on AI usage rather than work quality. He admitted that AI-written code "can be difficult to debug and is not consistently reliable."
How much time do Australian frontline workers lose to tech problems each month?
+
According to a SOTI report, Australian frontline workers in transport, logistics, healthcare, and emergency services lose approximately 13 hours every month, nearly two full working days, to device downtime, connectivity issues, and manual workarounds.
What is the AI Readiness Friction Framework?
+
Pearson's AI Readiness Friction Framework identifies six compounding frictions that prevent organisations from becoming truly AI-ready: pace, connection, capability, governance, experience, and skills. Only one of those six is a skills issue, the other five are management and organisational failures.
What percentage of AI implementation is people versus technology?
+
Successful AI implementation is roughly 80% people and 20% technology. Buying sophisticated AI software without investing equally in change management, training, and stable infrastructure is the primary reason most AI rollouts fail.
Which employees need the most AI training support?
+
Grant Thornton's survey of 950 C-suite leaders found that frontline employees (37%) and middle managers (30%) need the most support to implement AI effectively, yet most organisations direct training budgets upward to senior leadership instead.
Why do only 14% of graduates report high AI proficiency?
+
A six-country Pearson and AWS study found that 53% of employers struggle to find AI-skilled graduates and 78% of universities believe they're already meeting employer expectations, yet only 14% of graduates report high proficiency. Everyone thinks someone else is solving it, and nobody actually is.
Brett is a four-time founder (Darra Tyres, Gladfish, EzyTrac, Anaboo) and the operator behind AIOS, Anaboo's AI Operating System. He writes from inside the build, installing AI in his own businesses first and reporting back what actually moves the numbers. Based between Singapore, the UK and Australia.
