Our second MacDiarmid intern left us last week, leaving us with lots of great resources on artificial intelligence which, unlike quantum computing, the subject of Wayne Crump’s recently finished internship with us, is technology that is very much upon is.
As a taster for this reflection, I’ve included below a piece that Kyle led, co-authored by three of the advisors from the Chief Science Advisor Forum. The final version has been published on The Conversation. He also produced a useful annotated summary of resources (PDF, 128KB).
Check back to our interns page for his final report, which is currently being peer reviewed.
Thanks so much Kyle, for all your work and organisation of this massive topic, and again to the MacDiarmid Institute and all the researchers and policy folk who invested time in helping Kyle straddle the science-policy interface in an enormous topical area!
Artificial Intelligence is with us – what does it mean for our future wellbeing in Aotearoa New Zealand? A reflection on the recent reports from the Australian Council of Learned Academies and the Royal Society Te Apārangi.
Kyle Webster, Juliet Gerrard, Tahu Kukutai and Stuart McNaughton
- Kyle Webster has just completed an internship in the Office of the Prime Minister’s Chief Science Advisor, focussing on Artificial Intelligence
- Juliet Gerrard is a Professor at the University of Auckland and the Prime Minister’s Chief Science Advisor
- Tahu Kukutai (Ngāti Tiipa; Ngāti Maniapoto; Te Aupōuri) is a Professor at the University of Waikato and was a contributor to the ACOLA report
- Stuart McNaughton is a Professor at the University of Auckland and the Chief Science Advisor to the Ministry of Education
Over the last few days, the Australian Council of Learned Academies (ACOLA) and the Royal Society of New Zealand Te Apārangi have each released reports on artificial intelligence (AI). The first, ‘The effective and ethical development of artificial intelligence: An opportunity to improve our wellbeing’, is a detailed 200-page report that explores the topic in-depth and the second, ‘The age of Artificial Intelligence in Aotearoa’, provides an accessible summary, focussing on the opportunities that the so-called fourth industrial revolution might bring to Aotearoa New Zealand and the risks that accompany these opportunities.
The explicit focus on wellbeing provides a helpful framing to start a national kōrero about how we embrace these new technologies, and lead the conversation rather than passively accept advances from offshore. Aotearoa New Zealand has some advantages in its small size and strong legal frameworks to develop protocols for data security and intellectual property, and to nurture a diverse AI-savvy workforce. Placing equity at the centre of the conversation facilitates the adoption of AI technologies across education, government and industry in the context of public good. This framing mitigates the risk of importing biases from offshore algorithms, and allows us to identify areas where AI can benefit all New Zealanders, rather than a small international elite.
The report’s focus is on likely developments in AI with a twenty-year horizon – a useful timeline for policy development. In this window, specialist machine learning and narrow AI will enable us to begin to remove ‘the four Ds’ from our daily work (dirty, dull, difficult, dangerous), and focus instead on using machines to free people to use their originality, creativity and questioning abilities. It is unlikely that machines challenge these skills in this timeframe, and the arrival of artificial general intelligence, which could compete with these more human forms of cognition, is likely to be many decades away. By focussing on this nearer-term future, the ACOLA report is able to make more concrete recommendations on issues such as strategic investment, regulatory mechanisms including independent oversight and a design approach underpinned by principles of inclusion, and equity and respect for intrinsic human rights, whilst calling for a national strategy to bring these approaches together.
On the opportunities side, there are enormous potential economic gains in productivity resulting from AI-driven automation and optimisation, also signalled in the recent Government report: From the Knowledge Wave to the Digital Age from MBIE. Think orchard-based drones robotically applying herbicides in a highly targeted way, identifying weeds by AI driven vision recognition systems, then feeding all the data they gather into a central system that tracks and controls weed numbers across the orchard. Less herbicide use, better farm productivity, better environmental outcomes, and importantly, fewer people with jobs operating herbicide sprayers (at least in the way they do currently). In 2017, a Price Waterhouse Coopers analysis, put the expected contribution of AI technologies to the global economy by 2030 at $15.7 trillion USD. This is approximately the same size as the combined GDP of China and India in 2018.
The predicted gains in productivity are not without risks. Estimates of the proportion of jobs likely to be automated by AI vary wildly (Frey and Osbourne, OECD report), as do estimates of the degree to which those will be replaced by new jobs. Some of these new jobs are currently difficult to imagine – consider someone preparing for a career as a social media manager back in 2000, four years before Facebook was founded. Foresighting will therefore be key to build a resilient and adaptable workforce, with key skills in STEM subjects (especially maths) coupled with strong foundations in the humanities. There is a general consensus within the community of AI experts that the rate at which AI technologies may be disruptive to work and drive inequalities in income and employment opportunities will be a key concern, as AI technologies are implemented at scale. After all, even if the former orchard sprayer is able to find another job, either at the same orchard or another, they’re unlikely to get a direct financial payoff in the same way that the owners of the farm will. Multiplied thousands of different ways throughout New Zealand’s economy, these AI-driven changes threaten to widen inequities. Careful thinking needs to take place now to maximise the benefits for all. There is already work underway related to this issue at Otago’s Centre for AI and Public Policy (CAIPP) and this will need to feed into the policy space.
Economic effects are just one part of the story. The opportunities and issues raised in the reports range from lethal autonomous weapons, to threats for democratic decision making, to the disturbing gender imbalance emerging in the rapidly growing AI workforce. Additionally, the discussions fit into a wider body of recent work on AI relevant to Aotearoa New Zealand, including an analysis by CAIPP of the New Zealand Government’s use of predictive systems, establishment of the OECD principles on AI , and the NZ AI forum’s report on the state of the industry in Aotearoa New Zealand. A general follow up report and then a series industry specific analyses are due in the near future. The report identifies an urgent need for clear principles for use on the one hand, and state regulatory systems for socially accepted use and standards on the other. There is potential for inequities being exacerbated including through bias and misuses built into the technologies themselves.
The ACOLA report discusses the opportunities and issues for social sector policy, and rights and responsibilities under the Treaty of Waitangi. For example, in education, students need new knowledge and skills in preparation for, and to fully participate in, the near AI future. AI technologies should be developed to augment valued outcomes for students including critical thinking and social and emotional skills at all levels, and in a culturally inclusive context. More generally, the need to promote citizenship skills and knowledge for understanding and making informed judgements about uses – the very skills that the machines will not be able to assist with – will be vital. An important implication, consistent with the Government’s focus on Wellbeing, is the need to closely monitor societal impacts.
The ACOLA report is situated in a global context. It identifies areas where Aotearoa New Zealand could contribute to the global conversation, including in indigenous data sovereignty where tikanga-based approaches to these emerging technologies are being actively explored by groups such as Te Mana Raraunga and Te Hiku Media. Continued development of these knowledge systems could result in insights for other Indigenous peoples around the world. A Treaty-based approach to these technologies need to be further developed as a core part of our national strategy. As the ACOLA report notes “An AI strategy that places equity at its forefront will strengthen New Zealand’s international reputation in this arena and ensure that New Zealand is not left behind by some of the most important developments of the 21st century.”
In a global context, the report identifies a key opportunity for Aotearoa New Zealand and Australia to influence the direction of international governance of AI technologies. This will be through establishment and active participation in multinational groups involved in ethical and technical standard setting and coalitions of technologically advanced countries such as the D9 (of which NZ is a part).
The ACOLA report joins other groups, including the NZ AI forum, Te Kāhui Atamai Iahiko o Aotearoa, the Law Foundation and Callaghan Innovation White Paper in highlighting the role that nationally coordinated action can play in effective and fair deployment of these technologies.Aotearoa New Zealand has yet to formulate a national AI strategy. Perhaps it is time to open this conversation?
Overall, we see enormous potential for transformative opportunities in Aotearoa New Zealand that will improve wellbeing across all sectors of society, government, agriculture and industry. However, realising this potential will require placing a priority on addressing the substantial challenges inherent in ensuring a fair and equitable transition into an AI-enabled future, actively engaging, and ensuring benefit reaches those at risk of being excluded. The inclusive development of an AI-savvy workforce in Aotearoa New Zealand will also benefit from a coordinated approach to creating and supporting a diverse workforce with the critical skills, creativity and cultural competence to work in our local context.