Relief or Replacement? What AI Actually Means for Australia’s Teachers

by | Sep 1, 2026 | Education in the Age of AI

Skip to main content
< All Topics
Updated
ByThe Good

Any conversation about AI and jobs tends to default to a simple question — does it take yours? — and teaching is one of the few professions where the honest answer is closer to "it's arriving into a job that's already understaffed and overworked" than "it's coming for your role." That framing matters, because it changes the debate from replacement to something messier: whether AI eases a genuine crisis in the teaching workforce, or adds a new layer of pressure and risk to a profession already losing people faster than it can train them.

The workforce AI is actually landing in

The scale of the existing shortage is worth sitting with before anything else. The Australian Education Union's 2024 survey found 83% of the schools it surveyed were experiencing teacher shortages. The federal Department of Education projected a shortfall of 4,100 secondary teachers for 2025 alone, and modelling suggests Australia will need roughly 23,000 additional teachers by 2034 just to keep pace with population growth — before counting the roughly 34,000 needed to replace those approaching retirement. It isn't only a recruitment problem: AITSL data shows 35% of the current teaching workforce doesn't plan to stay until retirement, and around 20% of graduates leave within three years. Victoria alone is projecting a deficit of 2,052 teachers by 2030 — 1,675 of them secondary, concentrated in STEM subjects. On the OECD's TALIS 2024 survey, Australian lower-secondary teachers reported working an average of 46.5 hours a week — the third-longest working week in the OECD — with 34% reporting high stress "a lot," nearly double the OECD average of 19%. TAFE is arguably worse off again: a 2026 AEU survey of 1,696 TAFE teachers found widespread workload intensification and staff shortages, with 26% working more than 60 hours a week during term and 15–21 hours a week lost to administration alone.

There is a genuinely hopeful counter-trend worth holding alongside all of that. Undergraduate teaching applications rose 6.5% in 2026 to 15,302, with domestic offers up 6.3%, continuing a run of recovery after years of decline. Career changers are a growing share of new entrants — 52% of Teach For Australia's 2026 cohort came from other careers, and roughly 21% of early-career teachers nationally are now aged 40 or over. But that recovery is happening from a very low base, against a retention problem recruitment alone can't fix — NSW public school teaching vacancies did drop 61% between Term 3 2022 and Term 3 2025, real and welcome progress, but demand remains acute in growth corridors, western Sydney and low-SES areas, and rural and remote schools serving First Nations communities face the most compounded shortages of all.

The case that AI helps

The workload-relief argument is straightforward and, for many teachers, already lived experience: AI tools save real time on lesson planning, resource creation, communication and differentiating material for students at different levels, particularly around literacy support. Recognising that time constraints and a lack of mandated professional development were the real barrier, Microsoft's Elevate for Educators program launched in April 2026 to give Australian teachers and school leaders free training and practical resources for adopting AI safely and productively — a direct attempt to close the confidence gap rather than assume teachers would just work it out. Some edtech products go further still, using automated matching to speed up booking relief teachers and cut the hours school admin staff spend on manual phone calls, targeting one specific, chronic pain point directly. South Australia's EdChat, the government-built classroom AI assistant developed with Microsoft, was conceived explicitly to help teachers as well as students, rather than only managing the ways young people might misuse commercial chatbots. The pitch, in short, is that AI can absorb enough of the administrative and preparatory load that teachers get more time back for the parts of the job — direct instruction, relationships, judgement calls about individual students — that were the reason most of them entered the profession in the first place.

The case for caution

Set against that is the same TALIS data point from the primer article: 87% of teachers who already use AI flag it enabling students to misrepresent their own work as their top concern, meaning the workforce most enthusiastic about the tools is also the workforce most worried about what they're doing to the thing being taught. There's a cultural resistance running alongside the practical one, too. Some teachers are opting out altogether, unconvinced by the public criticism that's accompanied the AI boom, and one south-east Queensland state primary teacher described scaling back his own AI use because it gave the classroom a "cold, lifeless" feel he didn't want to bring into the room. That's not a data point about productivity; it's a data point about what teachers think the job actually is, and whether efficiency gains are worth a change to how the classroom feels day to day.

Deskilling is the quieter risk

Beneath the workload debate sits a subtler concern that gets less airtime: if AI increasingly drafts lesson plans, suggests differentiation strategies and pre-writes feedback comments, does the professional judgement that makes an experienced teacher valuable atrophy the same way concerns exist about students' critical thinking atrophying under heavy AI use? UTS's Professor Leslie Loble frames the underlying mechanism as being about how working memory converts into retained, durable knowledge — a description originally aimed at students, but one that applies just as easily to a teacher who stops actively planning a lesson and starts only editing what a model produced. Nobody has a clean answer to how much this matters in practice yet, partly because the tools are too new and partly because the profession is too stretched to run the kind of longitudinal research that would settle it.

The question worth sitting with

Teaching is a profession the labour market is actively begging for more people to join, at the exact moment a technology is arriving that could either free up the time needed to make the job survivable, or quietly hollow out the professional skill and personal connection that made people want to do it in the first place. Given the shortage is the more urgent, better-documented crisis, should schools be actively pushing AI adoption to buy staff back their time — or does a workforce already this stretched need to be especially careful about tools that could erode the judgement and connection that keep good teachers in the job at all?

Sources

Table of Contents