Queensland government leads the AI hype bubble
Queensland has promised a precise 25% cut to teacher red tape using AI. The private sector, with far more flexibility, still can't tell you what AI will deliver.
Queensland's Education Minister John-Paul Langbroek announced last week that the state government is cutting red tape for teachers by 25 per cent over four years, partly through AI-assisted administrative tools. The announcement was made with the confidence of someone who has already solved a problem they have not yet encountered.
The 25 per cent target has no measurement methodology attached to it
The media statement is worth reading carefully because of what it does not say. It commits to a 25 per cent reduction. It lists 37 actions under 9 commitments. It promises "innovative, integrated and responsive digital systems." What it does not do is explain how AI will achieve any specific component of that reduction, or what measurement methodology will be used to verify the number once claimed.
That 25 per cent figure deserves particular scrutiny. Red tape reduction is not like measuring rainfall. You cannot step outside and check it. Governments that commit to measurable administrative savings typically design the measurement at the same time as the intervention, which creates an obvious structural problem: the people responsible for delivering the reduction are also the people responsible for measuring whether they have. The document references "a range of tools to assess the ongoing impact" without specifying what those tools are.
Organisations with the least experience deploying AI are often the most precise about the benefits they expect from it.
Government AI adoption follows the Dunning-Kruger curve
The private sector is a useful reference point here. Australian businesses have been trying to extract real value from AI for several years now, and the honest verdict from most of them is that the technology is genuinely useful in narrow, well-defined tasks and genuinely disappointing when deployed as a broad productivity solution. The overhead of implementation, the cost of errors, the need for human review, the legal liability for automated decisions about students and staff — none of these disappear because the announcement sounds good.
As we noted when examining the federal government's own AI productivity claims, the mechanism by which AI lifts outcomes in complex institutional settings is rarely specified because it is rarely known. Everyone agrees the technology will help. Nobody can yet show the working.
This is the Dunning-Kruger shape of AI adoption in government. Organisations with the least experience deploying AI are often the most precise about the benefits they expect from it. The confidence is not evidence of competence. It is often evidence of the opposite: that nobody in the room has done it before and therefore nobody knows what the constraints are.
Private sector firms that have deployed AI in administrative functions at scale will tell you that the gains are real but uneven, that the implementation costs are routinely underestimated, that bureaucratic systems present particular challenges because the data is messy, the privacy requirements are strict, and the tolerance for errors is low. Schools are not software companies. The same teacher who cannot currently find a free hour in their week to learn a new system is the one being promised that AI will give them their time back.
AI is doing rhetorical work the reforms don't need
There is also a reasonable question about what is actually being automated here. The red tape reduction plan lists things like simplifying incident recording, streamlining communications, and rationalising policies. Some of that is sensible administrative hygiene that has nothing to do with AI. Some of it genuinely might benefit from well-implemented tools. The risk is that AI becomes the rhetorical wrapper around a set of reforms that would be unremarkable if announced without it, and the technology absorbs both the credit when anything works and the blame when the 25 per cent target is not reached.
The underlying problem — teachers overwhelmed by administrative load, spending time on compliance rather than students — is real. Queensland schools have the same structural issue that nearly every public school system faces globally: the administrative demands on teachers have grown faster than any productivity tool has been able to address them. If this plan produces genuine, measurable relief for working teachers, it will matter. That outcome is worth wanting.
But wanting it is not the same as having a mechanism for it. Governments that promise specific AI-derived outcomes before the implementation exists are not pioneers. They are just earlier arrivals to the disappointment.
Sources
Queensland Department of Education — Red Tape Reduction Plan 2025–28
The Bearing — Australia's productivity will be boosted by AI, somehow…
Frequently Asked Questions
What is Queensland's plan to reduce teacher red tape?
The Queensland government has committed to cutting teacher administrative burden by 25 per cent over four years through a plan listing 37 actions across 9 commitments, partly using AI-assisted tools. The plan covers things like simplifying incident recording, streamlining communications, and rationalising policies — many of which are standard administrative reforms that do not require AI.
Why is it hard to measure a 25 per cent reduction in red tape?
Unlike concrete outputs such as revenue or rainfall, administrative burden has no universally agreed baseline or unit. The Queensland plan references 'a range of tools to assess the ongoing impact' without specifying what they are, and the people responsible for delivering the reduction are also responsible for measuring it — a structural conflict that makes the target effectively self-reported.
Is AI actually useful for reducing administrative load in schools?
AI has shown real but uneven gains in narrow, well-defined administrative tasks in the private sector. In school settings, the challenges are more acute: data is often messy, privacy requirements are strict, tolerance for errors is low, and automated decisions about students or staff carry legal liability. Implementation costs in comparable bureaucratic environments are routinely underestimated.
Why do governments announce specific AI productivity targets if the technology is unproven?
Organisations with the least experience deploying AI tend to be the most precise about the benefits they expect from it — a pattern consistent with the Dunning-Kruger effect. Specific numbers signal ambition and attract attention, but when no mechanism exists to achieve them the precision is rhetorical rather than analytical.
What happens when government AI projects don't hit their targets?
When AI is used as a rhetorical wrapper around reforms that would be unremarkable without it, the technology tends to absorb both the credit when things improve and the blame when specific targets are missed. The underlying reforms — which may be sound — become harder to evaluate on their own merits.