Luke Munn provides a critical analysis of the current paradigms of artificial intelligence (AI) development and offer a framework for a decolonial AI. The author argues that existing AI paradigms reproduce and reinforce coloniality and its attendant inequalities. To overcome this, he proposes a framework based on Indigenous concepts from Aotearoa (New Zealand), which offers a distinct set of principles and priorities that challenge Western technocratic norms. The framework is centered on five tests that prioritize human dignity, communal integrity, and ecological sustainability. The author suggests that the application of these tests can guide the design and development of AI products in a way that is more inclusive, thoughtful, and attentive to life in its various forms.
The author identifies two distinct pathways for applying their framework. The first pathway is designing, which involves applying the principles and priorities of the Five Tests to the development of AI products that are currently in progress. This involves asking questions about how these products can respect the sacred, preserve or enhance life force, and reconcile negative impacts in acceptable ways. The author suggests that iterative versions of software can be developed by engaging genuinely with these questions and resolving them through code, architectures, interfaces, and affordances. The second pathway is decolonizing, which involves a deeper and more sustained confrontation with current AI regimes. This pathway involves challenging generic, universalizing frames, stressing the connection and interdependence of human and ecological well-being, and carefully considering potential impacts and developing ways to mitigate them or redress them to satisfy involved parties.
The author’s framework challenges current AI paradigms and practices by raising questions about what data-driven technology should be doing, how it can be designed in ways that are more inclusive, communal, and sustainable, and what values and norms should be used to judge the success of a particular technology. He suggests that these questions are epistemological, cultural and historical, and social in nature. And he argues that understanding and undoing systems of inequality that have been formalized and fossilized over time is a massive undertaking that demands a long-term project that prioritizes social justice.
In broader philosophical terms, this paper raises questions about the relationship between technology and power, the role of knowledge systems in shaping our understanding of the world, and the importance of Indigenous perspectives in challenging dominant paradigms. The authors’ framework challenges the Western-centric assumptions that underpin current AI paradigms and highlights the importance of recognizing and respecting diverse perspectives and ways of knowing.
Future research could explore the practical implications of the Five Tests framework and how it can be applied in different contexts. It could also examine the ways in which Indigenous perspectives can challenge dominant paradigms in other fields, such as philosophy, politics, and economics. Additionally, research could explore the potential for cross-cultural collaboration in the development of AI and other technologies, and how this collaboration can facilitate the recognition and respect of diverse perspectives and knowledge systems. Finally, research could explore the broader implications of the authors’ framework for the relationship between technology and power and the potential for decolonial approaches to reshape our understanding of the role of technology in society.
Abstract
As AI technologies are increasingly deployed in work, welfare, healthcare, and other domains, there is a growing realization not only of their power but of their problems. AI has the capacity to reinforce historical injustice, to amplify labor precarity, and to cement forms of racial and gendered inequality. An alternate set of values, paradigms, and priorities are urgently needed. How might we design and evaluate AI from an indigenous perspective? This article draws upon the five Tests developed by Māori scholar Sir Hirini Moko Mead. This framework, informed by Māori knowledge and concepts, provides a method for assessing contentious issues and developing a Māori position. This paper takes up these tests, considers how each test might be applied to data-driven systems, and provides a number of concrete examples. This intervention challenges the priorities that currently underpin contemporary AI technologies but also offers a rubric for designing and evaluating AI according to an indigenous knowledge system.
The five tests: designing and evaluating AI according to indigenous Māori principles