At first glance, it has some good-sounding words – but it ignores the real issues we should be worried about.
The paper has a big focus on using AI to boost economic growth and productivity. It’s true, we will be much more efficient at our jobs as AI continues to develop. But to what end? It’s not clear that we will work any less or that our jobs will be more fulfilling.
There’s also no guarantee that the cash from this economic boost will stay in New Zealand. The major players in AI – such as OpenAI and Microsoft – are based overseas and don’t have a good track record of paying taxes here.
New AI regulation is all but ruled out – on the basis it will stifle innovation, harm productivity, and be too hard to design. That scares me. It’s really easy to misuse AI technology in a way that causes real harm to people – and our current laws aren’t designed to tackle this.
Issues such as data privacy, political polarisation, and inequities in service delivery (to name just a few) will escalate without legislative oversight.
Sure, we don’t know exactly where AI will be in 5 or 10 years – but I think we can be proactive in requiring any AI tool to be used in a way that respects privacy, tests for bias, and can explain its decision-making. We definitely can’t afford to – as Collins suggests – rely on the industry to develop its own voluntary guidance and codes. AI companies have a bad habit of putting profit ahead of social good.
I’m not the only one who is concerned. A 2023 report referenced in this Cabinet paper shows 72% of New Zealanders are very or extremely concerned about unregulated AI. A similar number are concerned AI will be used maliciously or have unintended and harmful consequences. I’m proud Kiwis are critical of AI – so why is this Government not listening?
It’s not even like we would need to start from scratch. The European Union and United States, as different as they are, have both made big steps towards regulating AI. The EU’s world-first AI Act, which came into force this week, uses risk categories that place more stringent requirements on the riskiest applications of AI.
The Biden Administration in the US has used executive orders and developed a “Blueprint for an AI Bill of Rights” that sets minimum standards for AI safety and citizen rights in automated decision-making. We don’t have to reinvent the wheel – we just need to look overseas for inspiration to adapt to our specific context.
One part of this context – which the Cabinet paper is rather quiet on – is how the effects of AI will intersect with the Treaty of Waitangi.
Much has been written about the potential harm of AI for Māori – and on data/algorithmic sovereignty – but the paper simply considers that AI will bring “substantial improvements” in services for Māori. This feels incredibly naïve.
Still, at least the Cabinet paper seems to have been written by a human.
Back in March, Collins suggested such papers should be written with the help of AI.
Maybe she has changed her mind – or maybe Mbie, which drafted the paper, is still sticking to the AI ban it enacted in 2023, despite the minister’s enthusiasm for its adoption in the public sector.
