The Imperative of Aboriginal Data Sovereignty: Shaping the Future of AI

From the flowing streets of Sydney to the remoteness of the Australian bush, the interconnectivity of the modern world is a testament to our collective progress. Today, the digital age grants us the ability to send and receive information at unprecedented speeds, shaping economies, and politics, and actively forming subcultures. But while these advancements have provided many benefits, they also raise pressing concerns about the representation and autonomy of marginalized groups, particularly First Nations communities.

Data sovereignty, in essence, is the idea that data is subject to the laws and governance structures within the nation it is collected. As it pertains to First Nations populations, Aboriginal data sovereignty asserts the right of First Nations communities to own, control, access, and possess data that is about them. Such sovereignty allows these communities to tell their own stories, shape their own futures, and preserve their rich heritage.

“Creating such a dataset is a Pemulwuyaen (Herculaen) level task”

The conversation about Aboriginal data sovereignty comes into sharper focus when we turn our attention to the rapidly evolving field of artificial intelligence (AI), particularly large language models such as OpenAI’s GPT-4.

As these AI models become increasingly sophisticated, the datasets used to train them become crucial in determining the outputs they produce. These models learn from vast amounts of text data from the internet, encompassing everything from the scientific literature to social media posts. The crucial issue here is the lack of inclusion of Aboriginal and Torres Strait Islander narratives, voices, and knowledge systems in these databases.

In a world where AI plays a role in everything from generating text to informing policy decisions, the dearth of Aboriginal and Torres Strait Islander perspectives is questionable. Aboriginal and Torres Strait Islander communities possess unique cultures, languages, and histories that are not accurately or sufficiently represented in mainstream digital platforms and databases, this is changing with my people having great steps to digitalise language and culture but inclusion and creation of own data sets are vital

The absence of an Aboriginal and Torres Strait Islander cultural dataset in the training corpus of AI systems effectively renders these communities invisible to the AI, perpetuating historical patterns of exclusion. It also raises the risk of AI systems generating content that reinforces harmful stereotypes or misrepresents Indigenous cultures and identities.

The solution, though complex, begins with acknowledging the need for First Nations data sovereignty. A culturally inclusive dataset would incorporate Aboriginal texts, narratives, and languages, ensuring these AI models have a holistic understanding of the world’s cultures.

Creating such a dataset is a Pemulwuyaen (herculaen) level task. It requires collaboration between AI developers, anthropologists, linguists, and most importantly, Aboriginal and Torres Strait Islander communities themselves. Ethical guidelines should be implemented to respect the autonomy and sacred knowledge of these communities, ensuring data is gathered and used in a manner that is consensual, respectful, and beneficial.

Importantly, this process should not just be a matter of token representation. It must aim to incorporate Aboriginal and Torres Strait Islander knowledge systems into the very fabric of AI technology, honouring their unique perspectives and complex socio-cultural realities. This is not merely a question of technological advancement; it is a step towards a more inclusive, equitable digital future.

‘In a world where AI plays a role in everything from generating text to informing policy decisions, the dearth of Aboriginal and Torres Strait Islander perspectives is questionable’

This approach would not only enrich the AI’s understanding of human diversity but also empower Indigenous communities. It could contribute to preserving their languages and cultures, amplifying their narratives, and asserting their presence in the digital world. It is a chance to challenge the monolithic data hegemony, allowing for a myriad of voices to inform and shape AI technology.

The path to Aboriginal data sovereignty is neither straightforward nor easy. It will require the challenging of entrenched norms, the redrawing of ethical boundaries, and the willingness to listen and learn from communities too often pushed to the periphery. However, as we hurtle further into the digital age, it’s clear this is a path we must not only walk but also pave for future generations. In so doing, we ensure the AI of tomorrow represents all the world's people, not just a select few.