By copying the HTML below, you will be adhering to all our guidelines.

By Scott Alterator, Alexia Maddox, Clare Southerton and Stefan Schutt

Although it may not appear obvious, there’s a strong connection between teachers’ use of AI, the needs of First People’s children, and the findings of the Yoorrook Justice Commission.

Treaty in Victoria, a New Context

The Yoorrook Justice Commission has delivered comprehensive recommendations spanning education, health, criminal justice, and child protection—addressing systemic barriers faced by First Peoples in Victoria. The Commission produced 100 recommendations aimed at major institutional reform, with education findings revealing persistent inequities. First Peoples’ children remain more likely to be assessed as developmentally vulnerable, less likely to achieve top literacy and numeracy bands, and less likely to complete Year 12—outcomes that have shown minimal improvement over the past decade.

These findings underscore the urgent need for systemic change in how schools and teachers engage with Indigenous knowledge and support Indigenous students.

Against this backdrop, the emergence of generative AI (GenAI) in educational practice presents both opportunities and significant risks.

The Professional Challenge

The Australian Curriculum mandates the incorporation of Aboriginal and Torres Strait Islander histories and cultures across all learning areas. Yet, research indicates that over half of teachers above 35 years of age report feeling unprepared to teach First Nations content effectively. Only 11 percent of teachers feel confident to incorporate Indigenous perspectives in their classrooms. Educators across all career stages express concerns about cultural competence, fear of perpetuating stereotypes, and insufficient training in Indigenous pedagogies.

GenAI platforms promise efficient access to resources  and lesson planning support. However, their limitations in handling Indigenous knowledge raise serious professional and ethical concerns.

GenAI and Indigenous Resource Selection: Systematic Failings and Teacher Responsibility

Our research examined what could happen when teachers use ChatGPT and Microsoft CoPilot to find classroom resources about Aboriginal and Torres Strait Islander histories and cultures. We tested the models using a range of prompts that progressed to more complex requests, mimicking different levels of teacher competence with Indigenous protocols. Our findings revealed systematic failures when these platforms engage with Indigenous content, with their outputs including fabricated information, a perpetuation of racist perspectives and a lack of scaffolding in relation to potentially sensitive topics.

We used an adapted version of the App Walkthrough method to examine the GenAI outputs against two evidence-based frameworks offering practical guidance for educators navigating Indigenous content selection. Each offers a starting point grounded in Indigenist Standpoint Pedagogy, requiring educators to critically examine their relationship with knowledge production and their positionality within colonial systems.

Yarns Framework

The YARNS Framework provides structured questions examining authorship, representation, and cultural sensitivity. It guides educators to consider: Who created this resource? Whose voices are centred? What representations are being made? Does this material demonstrate respect for Indigenous knowledge systems?

The AIATSIS Guide to Evaluating and Selecting Education Resources

The AIATSIS Guide to Evaluating and Selecting Education Resources establishes a foundation in relational decolonisation. It begins with educator self-reflection—examining one’s own position, knowledge limitations, and responsibilities. The Guide then addresses content integrity, Indigenous authorship, cultural sensitivity, and proper attribution.

When we examined GenAI outputs against these frameworks, we found they consistently failed to meet basic requirements for acknowledging authorship of resources or respect for cultural protocols. The AI platforms repeatedly fabricated content. This included citing phrases from a non-existent First Nations language as part of an Acknowledgement of Country. It also included generating a list of fake documentaries as resources for a lesson. When challenged, ChatGPT defended its outputs before eventually acknowledging that the documentaries didn’t exist. We are still awaiting an apology.

The platforms also produced essentialising and deficit-based responses. We requested help to design a Year 9 science lesson and were encouraged to engage activities such as ‘comparing Indigenous perspectives with scientific explanations’. This positions Indigenous knowledge in opposition to science rather than recognising it as a valid knowledge system.

The platforms were not encouraging of teacher reflection. Prompts were not corrected in line with known errors, nor did they encourage a moment of pause for teachers to reflect on what the questions or outputs might reveal about the content or the nature of the request.

The Fundamental Incompatibility

The core issue extends beyond technical limitations. Indigenous knowledge is inherently relational – connected to Country, community, and guided by protocol. It is shared within relationships characterised by reciprocity and responsibility.

In contrast, GenAI operates through decontextualisation. It extracts information from sources, strips away context, and reassembles content based on probabilistic patterns. This process is antithetical to the relational nature of Indigenous knowledge systems. When AI platforms generate content about Indigenous topics, they present knowledge as neutral, context-free, and universally accessible. This approach perpetuates colonial knowledge structures that position Indigenous knowledge as inferior and Indigenous people as objects of study rather than authorities.

Professional Development Priorities

Yoorrook’s education recommendations emphasise the need for systemic change, including improved teacher education, culturally safe learning environments, and adequate resourcing. These priorities address the challenges revealed in our AI platform testing. Further, our research confirms the need for critical self-reflection as a foundation for teachers undertaking this work.

The YARNS framework and AIATSIS Guide provide structured approaches to this work, but they function as supports for professional judgement, not replacements for it.

Implications for Practice

The convergence of curriculum requirements, GenAI accessibility and promotion, and educator uncertainty creates significant risk. Well-intentioned teachers may inadvertently perpetuate the very colonial knowledge structures that decades of Indigenous scholarship and activism have worked to dismantle.

The frameworks developed by Indigenous educators offer pathways forward. They require time, reflection, and genuine engagement—resources that cannot be substituted by AI-generated shortcuts.

As educational institutions respond to Yoorrook’s recommendations, the role of appropriate resource selection and pedagogical approaches becomes increasingly critical. GenAI, in its current form, cannot support the relational, culturally responsive practice necessary for effective Indigenous education.

It’s up to education to meet the complexity with serious commitment

The Yoorrook inquiry has documented systemic failures and provided clear recommendations for reform. Implementation requires sustained commitment across all levels of education.

For individual educators, this means exercising professional judgment informed by Indigenous-led frameworks. It means recognising that some aspects of teaching cannot be automated or simplified without doing harm.

The tools exist. The knowledge is available. The question is whether educational practice will prioritise genuine cultural responsiveness over convenience. Professional integrity demands nothing less.

""

From left: Scott Alterator, Alexia Maddox, Clare Southerton and Stefan Schutt are the co-authors of Re-authorising Colonialism or re-authorising Indigenous perspectives: The Risks of Generative AI in Indigenous Education contexts.  Scott Alterator is an associate professor, Indigenous Education in the School of Education, La Trobe University. Alexia Maddox is a senior lecturer in Pedagogy and Education Futures in the School of Education, La Trobe University. Clare Southerton is a senior lecturer in Digital Education and Society, in the School of Education, Culture and Society, Monash University. Stefan Schutt is a senior lecturer, Digital learning Design in the School of Education, La Trobe University. 

Illustrations from Report to the Yoorrook Justice Commission from the First Peoples’ Assembly of Victoria

This article was originally published on EduResearch Matters. Read the original article."AARE"