Introduction
Generative AI tools, particularly large language models (LLMs) like ChatGPT and Claude, are already becoming embedded in teaching and learning practices, and we’ll likely see more of this as the technology and our familiarity with it develop. In this post, I’ll refer to these generative tools as “AI”. This shows up in many ways: teaching activities that model interactions with Copilot, assessments that encourage reflection on the use of AI tools, and the many ways students now use these technologies to support their learning.
Alongside this AI development is the ongoing process of decolonising curricula, with a critical examination of power structures in knowledge production, a questioning of whose knowledge is centred, and a rethinking of systems that preserve a Europe-centred lens. But how do these two conversations intersect?
When first considering the relationship between AI and decolonisation, the connection wasn’t immediately obvious. For me, the most intuitive link is that AI systems reflect the data they are trained on, meaning existing biases in knowledge production can be reproduced in AI outputs. This raises a deeper set of questions: how might AI technologies reinforce existing knowledge hierarchies, and how might they also help us see those hierarchies more clearly?
In this post, I reflect on that learning journey and explore one way we can engage with AI in the ongoing work of decolonising our curricula. Throughout this post, I use Global North and Global South as shorthand for long-standing global inequalities in knowledge production between regions, though I acknowledge that the labels encompass a range of countries with very different histories, economies, and cultural contexts.
Unequal inputs, unequal outputs
Let’s start with the most visible source of bias in AI systems: the data they are trained on. Colonial systems have historically controlled knowledge production, and the legacy of that control persists today, where the perspectives and ways of knowing that are foregrounded are largely those developed in Global North institutions and regions, while those of the Global South are sidelined, suppressed or erased1. These inequalities extend beyond training data, as access to AI technologies is also uneven, shaped by disparities based on gender, race, geography, income and societal factors2.
Because AI systems are largely trained on English-language and Global North sources, their outputs often reproduce dominant academic perspectives aligned with Western intellectual traditions rather than plural knowledge systems3. Without a critical lens, relying solely on AI outputs for learning risks presenting a skewed, incomplete picture of knowledge and reinforcing existing knowledge hierarchies.
Side note (could be presented in a box titled ‘Hidden labour behind AI’ at the side?): AI development also raises questions about labour and resource extraction that echo colonial dynamics. As Jones (2021) puts it, “the magic of machine learning is the grind of data labelling.”4 Many of the workers who label data or moderate content for AI systems are based in the Global South and work in precarious conditions, while the economic benefits of AI development are concentrated in large technology companies in the Global North, which hold disproportionate economic and political power over Global South economies3.
AI tools are often presented as neutral technologies that simply process information. In reality, like any technology, they reflect the systems and knowledge traditions that produced them. From a decolonial perspective, this matters because it means AI tools don’t simply deliver neutral information; they reproduce particular knowledge traditions, amplifying some perspectives while sidelining others. This is where AI becomes particularly interesting as a tool for decolonial teaching.
Side note (could be presented in a box titled ‘How Global South communities are shaping alternative AI approaches’ at the side?): While much discussion of AI focuses on extractive practices, many scholars and communities in the Global South are actively shaping alternative approaches to AI technology and knowledge production. These include efforts to rethink AI governance in regions such as Sub-Saharan Africa2, Indigenous communities using AI to curate archives and oral histories on their own terms5, and community-led data sovereignty initiatives where local organisations build their own infrastructure and embed Indigenous knowledge into data systems for challenges such as climate resilience6. AI does not have to reproduce colonial knowledge systems if we rethink who builds it and whose knowledge shapes it.
The opportunity of AI’s warped mirror
These limitations present an opportunity for learning and reflection. Used uncritically, AI could easily reproduce the same narrow perspectives that decolonising work seeks to challenge. However, by treating AI less as an all-knowing oracle and more as a warped mirror reflecting existing knowledge hierarchies, we have the opportunity to treat AI outputs as data for reflection, where the value lies not just in what AI produces but in how we interrogate those outputs. When approached with a critical lens, AI becomes a scaffold for developing evaluative skills rather than an authority that replaces human judgment.
To explore this idea yourselves, try submitting the following prompts to a generative AI system:
Prompt 1: Who are the foundational thinkers in [your subject]?
Then, consider the response:
- Which scholars are mentioned?
- Which regions are represented?
- Are Indigenous or non-Western perspectives included?
Prompt 2: Suggest five key readings on [a discipline-specific subject area].
Consider the response:
- What universities do the authors come from?
- What regions of the world are represented?
- What’s missing here?
In many cases, these responses will reveal a concentration of scholars from particular regions, institutions, or intellectual traditions. This doesn’t simply reflect the inner workings of the AI model itself, but the broader patterns of knowledge production that shaped the data it was trained on.
Decolonising education is not simply about diversifying curriculum content, but about helping students question how knowledge is produced, shared, and valued. AI systems can be a valuable tool in building those critical muscles.
These reflections can easily translate into classroom activities, such as asking students to:
- critique AI-generated reading lists
- analyse representation in AI responses
- investigate missing perspectives
- design alternative prompts which result in more representative outputs
Closing
There is little doubt that AI will continue to shape teaching and learning in the coming years. Future advances will depend not only on increases in computational power but also on how we use these tools to help students question how knowledge is produced and engage critically with AI rather than accepting its outputs at face value. The goal is not to reject AI tools, but to use them critically as prompts for reflection, helping students question assumptions and engage with a broader range of ways of knowing.
1. Schöpf, C.M., 2020. Conversations on the Global South–The coloniality of global knowledge production: Theorizing the mechanisms of academic dependency. Social Transformations Journal of the Global South, 8(2), p.2.
2. Ayana, G., Dese, K., Daba Nemomssa, H., Habtamu, B., Mellado, B., Badu, K., Yamba, E., Faye, S.L., Ondua, M., Nsagha, D. and Nkweteyim, D., 2024. Decolonizing global AI governance: assessment of the state of decolonized AI governance in Sub-Saharan Africa. Royal Society Open Science, 11(8).
3. Muldoon, J. and Wu, B.A., 2023. Artificial intelligence in the colonial matrix of power. Philosophy & Technology, 36(4), p.80.
4. Jones, P., 2021. Work without the worker: Labour in the age of platform capitalism. Verso Books.
5. Robert X (2024) AI’s power in decolonizing Global South histories. Available at: https://aicompetence.org/ais-power-in-decolonizing-global-south-histories/ (Accessed: 11 March 2026).
6. Bhanye, J., 2025. Artificial Intelligence (AI) and the New Colonialism of Climate Data in the Global South.
Many thanks to all of our collaborators for taking the time to contribute to this series.
View the other posts in this series here: https://bilt.online/category/decolonising/decolonising-ai/
Leave a Reply