A global review of 271 studies has found Australia has little local evidence to guide decisions about generative AI in schools, as researchers call for more research into how the technology affects student learning and development.
Researchers from the University of Sydney and the Barker Institute analysed studies examining the impact of generative artificial intelligence (GenAI) on young people from pre-kindergarten through to Year 12 across more than 45 countries.
Only four of the 271 studies were conducted in Australia.
The researchers say the limited Australian evidence base is significant as schools, educators and policymakers make decisions about the use of GenAI in education.
The review found growing evidence that GenAI can increase student engagement, motivation and confidence, assist students to complete learning tasks and improve the quality of their work.
However, there are major gaps in understanding its effects on deeper learning, reasoning, self-regulation and students’ longer-term development.
The research also found that outcomes vary according to how GenAI is used.
Five interconnected factors shape the impact of the technology, according to the review: the learner, the GenAI tool, the learning task, the social arrangements around learning, and the wider cultural and institutional context.
Students with stronger prior knowledge, for example, can be better equipped to question and evaluate AI-generated responses, while less experienced students may be more likely to accept answers at face value.
The design of the task also matters.
GenAI can support learning when it is used to practise skills, explore ideas, engage in reasoning or reflect on understanding. However, it is less likely to support learning when it performs the thinking students are expected to develop themselves.
Teacher involvement was identified as a critical factor.
Some of the strongest outcomes identified in the research occurred when GenAI was integrated with teacher guidance, combining personalised assistance and immediate feedback with professional judgement, subject expertise and human relationships.
Professor Lina Markauskaite, a Learning Scientist in the University of Sydney’s School of Education and Social Work, said the findings did not support a simple conclusion that GenAI was either beneficial or harmful to learning.
“What we found is not a simple story of GenAI being good or bad for learning,” Professor Markauskaite said.
“The evidence suggests outcomes depend on how these tools are used, what students are asked to do, and the support they receive from teachers.”
The review also found that research has largely focused on academic performance, engagement and self-regulated learning.
Less attention has been given to the broader social, civic and ethical capabilities schools seek to develop, including complex problem solving, ethical and intercultural understanding, personal and social capability, curiosity, courage and compassion.
Dr Natasha Arthars, a researcher in the School of Education and Social Work, said students were using GenAI in different ways, challenging assumptions that its primary purpose was to bypass learning.
“While the technology can certainly make that easier, that is only one possible outcome,” she said.
The researchers found some students use GenAI to generate answers, while others use it to test ideas, seek feedback, explore different perspectives and deepen their understanding.
For some students, the technology can also lower the barrier to seeking help, particularly when they may be uncomfortable approaching a teacher or speaking up in class.
Dr Hongzhi (Veronica) Yang, Senior Lecturer in the School of Education and Social Work and a core member of the Centre for Artificial Intelligence, Trust and Governance, said guidance should recognise that students at different stages of development may require different forms of support.
Professor Danny Liu, Professor of Educational Technologies and co-chair of the University’s AI in Education Working Group, said schools should move beyond debates about simply banning or embracing GenAI.
“The strongest message from this review is that GenAI works best when it enhances human learning rather than replaces it,” Professor Liu said.
The researchers say Australia needs more school-based research to establish what approaches to GenAI work best for Australian students and under what conditions.
The review suggests the evidence base will need to extend beyond immediate academic outcomes to consider how GenAI affects students’ reasoning, judgement, independence and broader development over time.