New guidance from Australia’s education evidence agency highlights what schools should consider when assessing AI tools, from learning benefits and student privacy to implementation costs.
Artificial intelligence (AI) may help teachers and students, but its impact on learning is not yet well understood. New guidance from the Australian Education Research Organisation (AERO), published on 14 October, outlines considerations for school leaders and teachers as they decide whether, when and how to use AI in education.
AERO says the educational value of AI depends partly on how tools are designed and used, including whether they support the thinking required for learning or replace it. Its guidance identifies five considerations: evidence-based teaching and learning, risks to learning, clear expectations for AI use, assessment, and culturally responsive practice.
For school leaders evaluating AI products, these findings raise a practical question: what should schools establish before committing to a new technology?
Read AERO’s guidance for school leaders and teachers.
1. What evidence supports its educational value?
AI products may promise personalised learning, greater engagement, improved student outcomes or reduced teacher workload. However, AERO says the evidence on AI’s impact on school learning remains limited, with much of the available research conducted overseas, over short periods or in secondary and higher education settings.
AERO also highlights a potential “performance paradox”: students may complete tasks more successfully with AI assistance but learn less in the process. The concern is whether students are acquiring and retaining knowledge, rather than simply producing a better immediate result.
School leaders should therefore ask vendors for evidence supporting their claims and consider whether it applies to the intended age group, subject and classroom context. Engagement statistics alone do not demonstrate improved learning.
2. How will student information be protected?
Schools need to understand what information an AI product collects, how it is used, who can access it and whether it may be used to train AI systems.
The Office of the Australian Information Commissioner (OAIC) recommends due diligence when organisations select commercially available AI products, including consideration of privacy and security risks, suitability for intended uses and appropriate human oversight.
Questions for vendors include how long data is retained, whether it is shared with third parties, what happens when a contract ends, and what protections apply to information entered by students and staff.
Schools should also check relevant privacy obligations, departmental requirements and internal policies before using tools with personal or sensitive information.
The OAIC’s guidance on privacy and commercially available AI products provides further information.
3. Is the tool appropriate for its intended users?
AI-generated responses can be inaccurate, misleading or biased. Students who lack the background knowledge to evaluate an answer may struggle to recognise errors or fabricated information.
School leaders should assess age suitability, accessibility, safeguards and the level of teacher oversight required. AERO advises greater caution when considering AI for primary-aged students, given the limited research involving younger learners and their developing capacity to evaluate AI outputs independently.
Leaders should also consider cultural responsiveness. AERO warns that AI systems may misrepresent First Nations knowledge, cultures and languages, reinforcing the importance of appropriate sources, community engagement and respect for Indigenous Cultural and Intellectual Property.
4. Will it support teaching, assessment and existing systems?
Before adopting a product, schools should consider how it fits with curriculum requirements, assessment practices, existing technology and staff capability.
AERO recommends clear expectations about when AI can be used and when students must demonstrate their own knowledge and thinking. It also highlights challenges in determining what students can do independently in take-home or unsupervised assessments.
Implementation decisions should account for staff training, technical support, integration and ongoing review, as well as how teachers will maintain responsibility for assessment judgements and feedback.
Australia’s Framework for Generative AI in Schools provides further guidance on responsible and ethical use.
5. What is the total cost, and how will success be measured?
Subscription fees may not reflect the full cost of adopting an AI product. Schools should consider implementation, training, support, renewals and the costs of changing or exiting a service.
Before committing funds, decision-makers should identify the problem the product is intended to solve and establish how its value will be assessed. Measures might include learning outcomes, teacher workload or improvements to a specific school process.
AERO’s policymakers explainer also identifies emerging quality-assurance work relevant to schools. The Safer Technologies 4 Schools initiative is piloting a Responsible AI Standard and Evaluation, while AERO and Education Services Australia are piloting national quality standards for education technology, including AI-enabled applications.
These initiatives may help inform future product evaluation, but schools should verify the status and applicability of any assessment before relying on it in procurement decisions.
Making an evidence-informed decision
AERO’s guidance does not endorse particular AI products or provide a universal procurement checklist. It does, however, emphasise that decisions should be guided by evidence about how students learn, with careful consideration of whether technology supports or undermines that process.
AI in schools: Considerations for school leaders and teachers