New Zealand’s innovative approach to public policy is focused on a broad range of outcomes to enhance the wellbeing of its citizens as opposed to taking the social investment approach of focusing only on fiscal liability.
Critical to its delivery is artificial intelligence, or AI, which enables the New Zealand government to both understand the challenge of delivering wellbeing and supporting the prioritisation of initiatives for wellbeing.
At an event hosted by The Mandarin and analytics leader SAS in Canberra on August 6, New Zealand’s journey with AI was put in the spotlight with speakers James Mansell and SAS Australia and New Zealand Director of Customer Advisory, Dominic Frost, sharing insights into outcomes, benefits and the lessons learned.
The New Zealand transformation
In the 1980s, Mansell explained, the New Zealand government’s reforms were focused on building efficiencies – including introducing competitive contracting, cost savings, winding back office duplication, and improving the ability to do accounting and financial management.
Recent reforms led by former Prime Minister Bill English, saw an increased focus on linking the use of data and analytics to focus on better investing in impact. But the focus was narrow and predominantly on fiscal liability outcomes.
This year with the introduction of the wellbeing budget in May, the focus has broadened to better include real outcomes for people. Language within the government is also changing with people no longer called liabilities but assets. “Social investment” is now “wellbeing”. And it’s not about micro-targeting but about “proportional universality”, according to Mansell.
AI is still at the core of supporting wellbeing. The Social Investment Agency now has core funding under the new government for data linkage and analytics, to support collective impact initiatives for wellbeing. Money is still flowing into AI, including $30 million for citizen centred analytics and automation in New Zealand’s tax agency. And a new frontline data sharing initiative is in the works to support better collaboration between agencies for citizen-focused outcomes.
The use of AI in New Zealand intends to move the work of government to be outcome focused. And its use will help build sophistication around questions asked. Rather than programs that use overly simple categories to target support such as “single mothers” – which focus on a couple of demographic categories – they can build models that can incorporate additional demographic, geographic, economic and other factors to target programs and services that support the varying needs of single mothers across New Zealand.
The transformation in New Zealand has not been without challenges. Identifying the role of AI, and the role of people to support it, has been important. Slowly building the case studies, expectations and network of advocates has also been important.
“The two mistakes you can make is trying an all-singing, all-dancing approach and the other problem is not being focused on the business value,” Mansell said. “Starting small and figuring out how to use this [AI] for genuine business value is the key.”
And having the right people to sell the value has been critical.
Defining and selling AI
Frost explained that AI is becoming more important to government and business is generally because of three factors: cheaper computing power, better algorithms and the fact that we capture data everywhere. But for internal teams to “sell” AI within government, there needs to be a clear understanding of what is being discussed.
“We’re not talking about machines replacing humans,” he said. “AI is a broad-brush term and that is one of the challenges. There are a range of things that can mean AI including natural language processing, machine learning, predictive modelling, computer vision, and even things like forecasting can fall within the definition of AI.”
AI can provide value to government in three areas: decision making, automation of dull and repetitive tasks that can be done faster and with more accuracy than humans, and in predictive modelling.
“One of the risks I see in this area is that there is a lot of experimentation going on versus success projects – people are trying things that are not core of what they do today,” Frost said. “People think the first AI project they have to take on is Google or do weird and wonderful things. But the biggest potential benefit in AI is the dull, difficult and dirty.”
Where AI is designated within the organisation is also an important consideration for the sales of AI. If it is seen as an ICT solution, the “internal sales team” are the IT team or data scientists. Their pitches become bogged down in the technical components of AI processes and models. And the focus is not on business outcomes and results.
A purely technical pitch to senior executives, Frost finds, rarely gets internal approval or momentum.
But embedding AI projects within business divisions that are at the frontline of delivering policy and services means the technical experts are working with subject matter experts to deliver business outcomes. And the pitch for AI becomes a “no-brainer”.
This was an important lesson for Mansell in his early work on AI in New Zealand.
“One of the most important things I was told was that I needed to get out and start talking to people,” he said. And Mansell learned to approach the use of AI with humility, framing insight as discovery for experts to interpret, rather than saying the model knows best.
Leading the AI transition in Australia
New Zealand still faces challenges with the use of AI, particularly in the area of data sharing. But framework, policies and prototyping continue to reduce the barriers. The challenges New Zealand has faced are also seen in Australia – just on a larger scale. Yet they are barriers that can be addressed.
“I see the same appetite in Australia and Australia is getting great outcomes in certain areas – but there is a bit more complexity here,” Frost said. “New Zealand benefited by being first out of the blocks and having pretty strong leadership. If you are going to do any change – it doesn’t matter if it is technology change or organisation change – if you have strong leadership, you can drive it through.”
Leadership, Mansell agreed, is critical in the AI formula for success to ensure outcomes are business and people focused.
“I see analytical insights and AI as a change leadership challenge,” he said. “An insight by definition tells you something you didn’t already know. And that confronts existing practice. So, generating insights is one thing. Turning insight into action is where the real challenge lay.”
