If AI is used correctly, it might finally be the answer to solving education inequality in Australia.
The generative and adaptive capacity of Artificial Intelligence are already no stranger to teachers in Australia. Our own surveys directed towards staff found that an overwhelming 97% of all teachers found AI generated content in submitted work by students, and no lack of complaint in an effort to identify and reprimand instances of academic plagiarism and misconduct.
However, our own team is now working with teachers to see if these tools can finally be the key to giving every student something they have never truly had access to before, a learning experience built entirely around them.
The traditional classroom model has not changed in any fundamental way since the industrial revolution, despite the massive improvements in technological outcomes. A teacher stands before a cohort of 20 to 30 students, delivering a curriculum bounded by school terms and fixed timelines. Beyond the sector’s uncertainty in facing AI is a clear need for more attention and funding seen through the increasing amounts of schools resorting to makeshift classroom “demountables” to account for rising student numbers and a lack of specialised subject teaching staff.
The inevitable outcome, as Alex Jenkins from the WA Data Science Innovation Hub notes, is that teachers are effectively forced to teach to the average. Students who fall behind rarely catch up. Students who are ahead are rarely pushed further. Anyone who strays from the standard is forced to undergo a curriculum not suited to their capacity, as best as teachers may try to help stragglers.
According to a 2022 Productivity Commission report, between 5 and 9 per cent of Australian students fail to meet year-level expectations in literacy or numeracy each year. The report itself identified targeted, one-on-one instruction as the most effective intervention, yet this is precisely what a single teacher with thirty students cannot consistently accomplish.
This is where AI can serve not as a threat to tried and tested education methods, but as a genuine opportunity for innovation. If AI educational assistants can work alongside every student, and adapt to their individual learning style, algorithms can fill gaps in prior knowledge before new concepts are introduced, and flag where additional support is needed.
For neurodivergent students in particular, the capacity to shift between visual, auditory and conceptual teaching methods on the fly could be transformative.
Teachers can finally be given the desperately needed time to give more tailored feedback and monitor students progress individually with these tools, being freed from the administrative weight of trying to individually track and respond to thirty students at once. This means that more dedicated support requiring human connection can be given to students attention must be given to the most. Already, the trialling of an online chatbot “EdChat” for schools and educators in South Australia is seeing some success and could be the blueprint in other states too.
The consideration of potential risks, particularly in a sector where information and lesson quality is paramount is similarly higher than ever. The current EdChat feeds off a limited version of GPT-4 (as of 2025) and any oversight in the data it’s trained on must ensure the potential for hallucination and polluted data are genuine concerns. Any sanctioned data AI tools can and do produce information that must not deviate from established curricula, and the potential for both students and educators to develop an overreliance on these tools is a genuine concern.
As with any technology introduced into schools, the question of who controls it, how it is monitored and what guardrails are in place matters enormously. Irrespectively the opportunity is real. Whether Australian schools have the will to embrace it remains the question worth asking.