There’s a comfortable story a lot of organisations tell themselves about AI skills. Something like: this is moving fast, but we’ll catch up, everyone will catch up, it always works out. The data coming out of Australia right now doesn’t support that story.
Put a few numbers side by side and the picture gets uncomfortable fast.
The four numbers that matter
500%. That’s the growth needed in Australia’s AI-capable workforce to meet demand by 2030, according to the Australian National AI Plan. Not 50%. Not 150%. Five hundred percent, and we are so behind! This isn’t a gap that closes with a few training courses, or webinars. It’s a structural shortfall.
86%. The share of companies naming AI literacy as the fastest-growing skill in the Australian workforce, per LinkedIn’s Skills on the Rise research. Employers already know this. This isn’t a case of businesses being slow to notice a trend. They’ve noticed. They’re just not sure what to do about it yet.
97%. The share of hiring managers who now expect new hires to have AI skills, according to Robert Half’s Australian employment survey. Read that number again. Ninety seven percent. AI fluency has quietly stopped being a nice-to-have on a resume and become a baseline expectation, almost overnight, and most candidates don’t seem to have registered the shift yet.
74.9%. Undergraduate satisfaction in the latest QILT Graduate Outcomes Survey, the lowest recorded since QILT began tracking in 2016. Graduates are telling us directly that they don’t feel prepared. That’s not a vague sentiment. It’s the lowest number on record.
Why these four numbers tell one consistent story
Individually, each of these stats is notable. Together, they describe a system that is misaligned at every level.
Employers know AI literacy matters (86%) and are already hiring on the assumption that candidates have it (97%). Universities are producing graduates who feel underprepared for what employers now expect (74.9% satisfaction, the lowest ever recorded). And the workforce overall needs to grow its AI capability five times over just to keep pace with demand by 2030 (500%).
Nobody in this picture is short on urgency. What’s missing is a shared way to measure the thing everyone is worried about.
Right now, if a company wants to know whether a candidate is genuinely AI fluent, they’re relying on self-reported claims on a resume, a line that says “proficient with AI tools,” which means nothing and can’t be checked. If a university wants to know whether its graduates are workforce-ready on this specific dimension, there’s no consistent benchmark to test against. Everyone is guessing, at exactly the moment guessing has become expensive.
We have a measurement problem, not a training or learning & development problem
It’s tempting to read these numbers and conclude the answer is more training. Run more workshops, add an AI module to the curriculum, send staff to a course. Training matters, but training without a way to verify what actually landed is just activity. You can run a hundred workshops and still not know whether the people who sat through them can reliably delegate the right task to AI, describe what they need clearly, judge whether the output is right, and take responsibility for the result.
That’s the actual gap. Not a lack of effort. A lack of a reliable, shared way to measure and verify AI fluency, the same way a language proficiency test measures whether someone can actually speak a language, not just whether they attended classes.
That’s the specific problem Gradstack’s AI Fluency Index, GAFI®, was built to solve. It gives employers and universities a single, verified way to score AI fluency, individually and at population scale, and to track how it improves over time. A benchmarked score that can help move the needle forward on all fronts and for everyone.