84% of Australian Workers Use AI. Only 7% Use It Well.

Eighty-four per cent of Australian workers now use at least one AI tool at work. That sounds like progress. Then you look at how well they use it: only 7 per cent qualify as advanced users. More than half — 54 per cent — are still beginners. The gap between those numbers, according to RMIT Online and Deloitte Access Economics, is worth $18.9 billion in unrealised economic growth.

The Beyond Prompting report, published in March 2026 and based on a survey of more than 2,000 Australian workers, doesn't measure whether people use AI. It measures whether they use it well. The distinction matters more than most business owners realise.

Here's the finding that should concern every business owner: Australian workers score twice as high on technical AI skills as they do on judgment skills. They can write prompts. They can generate output. What they can't do — reliably — is evaluate whether that output is correct.

The consequences are already showing up. A University of Melbourne and KPMG study found 56 per cent of Australian workers have made workplace mistakes using AI. Over half have presented AI-generated content as their own without verifying it. Among large Australian and New Zealand companies, 95 per cent have reported at least one AI-related incident in the past two years. Eighteen per cent lost between US$1 million and US$2 million per incident. Eight per cent lost more.

For a five-person trades office, the dollar figures are smaller but the pattern is identical. The scheduler who accepts AI-generated route plans without checking geography. The estimator who uses AI for quotes but doesn't verify material quantities. The bookkeeper who trusts Xero's AI categorisation without reviewing the classifications. Each of these shortcuts works most of the time. The rest creates rework, client complaints, and errors that compound until someone notices.

The RMIT data puts a dollar figure on the literacy gap at the individual level. Workers who advance from beginner to intermediate AI literacy earn approximately $7,000 more per year — a 6.2 per cent wage premium. Moving from beginner to advanced: nearly $11,000. Those wage premiums reflect productivity differences — more output, fewer errors, better decisions.

We wrote recently about the gap between AI adoption and integration — 69 per cent of Australian SMEs use AI, but only 12 per cent say it's core. The RMIT data explains part of the reason: most workers are still using AI at a beginner level. The tools are deployed. The skills to use them properly aren't.

The generational breakdown adds another layer. Seventy-six per cent of baby boomers are beginner-level AI users, compared to 43 per cent of millennials. But younger workers have a different problem: overconfidence. Twenty-one per cent of Gen Z workers overrate their own AI abilities, versus 8 per cent of boomers. One group doesn't use AI enough. The other trusts it too much. Both create risk.

Only 48 per cent of Australian workers receive any formal AI training from their employer. Among baby boomers, 52 per cent haven't had any in the past year. Nearly half of all workers — 49 per cent — are teaching themselves through trial and error. They're building prompt skills without building the judgment to know when the output is wrong.

The Australian Government, through the National AI Centre and TAFE NSW, is offering one million free AI scholarships — an online, self-paced microskill course built around the government's Guidance for AI Adoption. It covers responsible use, risk awareness, and practical application. It takes roughly a day to complete.

Three things this week. First, have every team member complete the free government AI course. It won't make them advanced users, but it will move most from beginner to informed — and start closing the judgment gap that costs you money. Second, pick one AI workflow your team already uses and add a verification step. Not bureaucracy — a 30-second check that the output is correct before it reaches a client. Third, ask your team what AI tools they're actually using and where they feel least confident. The RMIT data shows the biggest risk isn't AI itself. It's AI used without the skills to evaluate what it produces.

Key takeaways

Not sure where your team's AI skills stand?

A short call can identify the biggest AI literacy gaps in your business and the practical steps to close them. Book a call.