Late last week, the Office of the Minister of Science, Innovation and Technology proactively released a paper submitted to Cabinet addressing New Zealand's approach to working with artificial intelligence (AI).
The Cabinet paper notes that New Zealand has been uncharacteristically slow to embrace AI and identifies confidence and capability as key factors inhibiting the rate of domestic AI adoption, particularly compared to peer countries.
The proactive release of the paper appears to be aimed at giving an immediate boost to business confidence, by:
- Clarifying the likely regulatory direction that New Zealand intends to take (ie taking a "light touch", facilitative approach that avoids intervention except to address acute risks)
- Announcing plans to improve market capability through new cross-portfolio, government leadership initiatives.
New Zealand "behind its peers" on AI adoption
The paper comes from the office of Hon Judith Collins KC, who made a number of campaign promises relating to prioritising issues relating to AI, asserts that New Zealand is "behind its peers" in AI adoption, and points to a number of local survey results to support this view. In 2021, a Qrious survey noted that only 28% of businesses agreed they had a good understanding of the legal and ethical implications of AI. Two years later, a Datacom survey of 200 senior business leaders found that roughly half had yet to take up AI in their organisation, with some reporting no plans to do so.
The paper does not itself identify any overseas surveys that demonstrate New Zealand's perceived uptake deficit, however a 2024 Deloitte report noted that 79% of Australian business and technology leaders expect generative AI to drive substantial transformation within their organisations and industries in the next three years.
Next steps to address the challenges: the five domains
The paper proposes that Cabinet approves "decisions and actions" in five domains.
1. Setting a strategic approach
- The paper provides a clear and unambiguous statement of support for the increased use of AI in New Zealand. Businesses looking to invest in AI initiatives may find particular assurance in the explicit directions that regulation of AI should be "light-touch", "proportionate", and that regulatory intervention "should only be considered to unlock innovation or address acute risks". The proactive release of the paper seems aimed at communicating the message that businesses should no longer delay investing in AI for fear of future regulatory and compliance risks.
- The paper also explicitly discourages the development of a standalone "AI Act", favouring instead the use of relevant existing frameworks (such as those relating to privacy, consumer protection, and human rights) as these have the benefit of being technology neutral. Where required, more agile regulatory tools such as industry codes or technical standards could be used to address acute risks, rather than legislating on the basis of "speculative harm". This suggested approach contrasts with the ground-breaking approach taken by the European Union earlier this year, which introduced a new Artificial Intelligence Act that classifies AI systems according to the risks they pose, and bans the use of some of the highest-risk systems.
- The paper also affirms that the OECD AI Principles should "serve as a visible signal of our national approach and high level direction".
2. Enabling safe AI innovation in public services
- We recently published an article on the regulation of algorithm use within New Zealand government agencies, focusing on the overlapping, principle-led frameworks and initiatives that are in place to support government agencies effectively regulate their own use of algorithms. The continuation of this programme of work is identified in the paper as a strategic foundation for efforts to increase AI uptake. As we noted in our article, if AI was to be identified as the cause of any public service delivery failure, or be viewed as perpetuating unintended negative outcomes for individuals or groups, the consequences are potentially serious - trust and social licence are difficult to build and easy to lose.
- Specifically, the paper notes the work of the Government Chief Digital Officer (GCDO), endorsing the continuation of the current initiatives and referring to the potential development of a new assurance model for the GCDO's portfolios.
3. Harnessing AI in the New Zealand economy
- While the details are scant, this is the domain that includes the most immediate near-term deliverables. In it, Minister Collins confirms that she has asked the Ministry of Business, Innovation and Employment (MBIE) to develop and deliver "risk management-based guidance" on the use of AI for use by the private sector, as well as an "AI Roadmap". The purpose of the AI Roadmap appears to be to signal to private sector firms where the government will be focusing its innovation efforts and when, so that private firms are able to align their own efforts, optimising domestic productivity and export potential.
- MBIE is currently engaging with the sector to inform this work and develop buy-in. The paper recommends that Cabinet issues a direction to agencies to provide MBIE with inputs for this cross-portfolio work programme by the end of August 2024, with Minister Collins to update Cabinet on progress on this programme of work in September 2024.
4. Prioritising engagement on international rules
- While New Zealand has consistently engaged with global initiatives and communities relating to AI and its regulation in recent years, the rapid proliferation of such initiatives means that New Zealand now needs to establish an effective order of priority for these engagements going forward, in a manner that best supports New Zealand interests.
- To do so, the paper suggests inviting the Ministry of Foreign Affairs and Trade to assess which of the emerging and existing policy responses to AI are most beneficial for New Zealand to focus on.
5. Coordinating with work on national security
The paper acknowledges the work of the Department of the Prime Minister and Cabinet as the strategic coordination agency for emerging, critical, and sensitive technologies (ECST) affecting national security, and suggests that the AI and ECST workstreams should "seek to complement and coordinate with each other", due to the high likelihood of overlap.
Support for the approach and initial response
The paper notes that a Cross-Party AI Caucus met earlier in 2024 and that there is cross-party accord that the government should support the use of AI by government to improve services for New Zealand. Minister Collins also confirmed that she had met with the Data Iwi Leaders Group (DILG) and was pleased with the "ready uptake" of AI by the DILG, noting that "there seems to be a common acceptance that the danger for Māori in AI is to be left outside of its ability to bring about substantial improvements in the provision of services for Māori".
Early reaction in the media also reflects a relatively positive response from private sector stakeholders, however it remains to be seen the extent to which the light-touch regulatory approach endorsed in the paper will come under specific public or academic scrutiny. The paper itself notes that a 2023 “Internet Insights” survey by Verian found over 66% of New Zealanders were "extremely" or "very" concerned that AI may be used for malicious purposes, and also refers to a Xero survey of small businesses that showed 78% of New Zealand respondents were concerned that AI development and adoption was outpacing regulation.
While the latter finding is used in the paper to illustrate how mistrustful attitudes are holding back the beneficial diffusion of the technology in New Zealand, the paper does not make it clear how the Government intends to sell the public on its policy of boosting AI adoption with minimal regulation, when more than three quarters of the public are reportedly concerned that adoption is already outpacing regulation.
This article was co-written Andy Dysart (senior associate) and Renee Stiles (partner).
