The Australian Human Rights Commission (Commission) is pleased to provide its submission to the United Nations' Office of the Secretary-General’s Envoy on Technology in response to the Call for Papers on Global Artificial Intelligence (AI) Governance.
The Commission's submission will assist the High-Level Advisory Body on AI's (Advisory Body) approach to the global governance of AI.
AI has the potential to improve our lives in meaningful ways. However, if it is not developed and deployed safely, it can also threaten human rights.
Right to privacy
The right to privacy is a cornerstone human right that also underpins freedoms of association, thought and expression, and freedom from discrimination.
The operation of AI may facilitate and deepen privacy intrusions in new and concerning ways. AI products are trained on large data sets, which often include personal information – this incentivises a broad approach to collecting, storing, and processing vast amounts of data.
The collection of data to train AI products will only heighten existing issues surrounding data collection.
Algorithmic bias arises when an AI tool produces outputs that result in unfairness. Algorithmic bias can entrench unfairness and even result in unlawful discrimination.
For instance, a 2019 study uncovered that a clinical algorithm used by many hospitals in the United States to determine which patients required extra medical care produced racial bias.
The algorithm was trained on past data on healthcare spending, which reflects a trend whereby black patients have less income to spend on their healthcare as compared with white patients – a result of systemic wealth and income disparities.
The algorithm’s outputs, therefore reflected a discriminatory result of white patients requiring more medical care than black patients.
Algorithmic bias is a highly technical issue that the Commission has substantively engaged with in its Technical Paper.
Centring human rights in AI governance
It is of utmost importance that the Advisory Body have regard for human rights when considering the global governance of AI.
Any response must consider a combination of regulatory responses to the human rights risks associated with AI, in addition to non-regulatory measures.
Providing guidance on the use of AI will be especially important in ensuring ethical AI, and the Commission supports the development of practical advice - such as its AI in Banking HRIA Tool and Resource on AI and Discrimination in Insurance.
Please read our full submission on the importance of centring global AI governance on human rights to better understand the risks and benefits of ethical AI.