A Public Australian AI

May 8, 2024

If AI is going to be as consequential and transformative as is being predicted, then we need to build AI infrastructure that is made in Australia, with a public good/not-for-profit imperative.

For over a decade, we’ve been content to allow private, foreign owned digital platforms to slowly infiltrate our public services, community forums and democratic processes. This has included critical areas such as news dissemination, information sharing during disasters, community consultations, public service provision and even public trials.

Slowly, we transitioned our public communications infrastructure from publicly managed platforms to privately owned digital products and social media. Social media and digital platforms have created an atmosphere of distrust, with most Australians believing social media “causes more problems than it solves”.

The saturation of social media for our public services, and the distrust Australians have for it is worth noting for the development of AI. Already, the most dominant players in AI are shaping up to be only a small handful of overseas companies. A report from the CSIRO which looks at AI foundation models found that the vast majority of AI models are from the US (73%), followed by China at 15%, with the rest from the EU and other countries.

This is particularly concerning when we consider that AI will need to account for specifically local concerns and outputs – requiring local data sets, and local quality checkers. AI is meant to be reflective and representative of a particular region’s culture and information, enough so that its recommendations are understood and tailored to that region’s context. Why then, would we rely on foreign companies to decide this local context for us?