Credit: Dreamstime
Microsoft’s government affairs lead for New Zealand has suggested that the New Zealand government should consider developing practical implementation guidelines as a next step following the introduction of its new public sector algorithm charter.
It was revealed on 28 July that 21 government agencies had signed up to the Algorithm Charter for Aotearoa New Zealand, a new set of standards introduced by the government to guide the use of algorithms by public agencies.
The Algorithm Charter for Aotearoa New Zealand follows a recommendation by the Government Chief Data Steward and Chief Digital Officer, who said in a 2018 report that the safe and effective use of operational algorithms required greater consistency across government.
Broadly, the charter is designed to commit signatory agencies to a range of measures, including explaining how decisions are informed by algorithms and embedding a Te Ao Māori perspective in the development and use of algorithms.
“Building on previous artificial intelligence (AI) research by the New Zealand Law Society, the charter emphasises that more complex algorithms can be used to support human decision-making, helping the government to better understand New Zealand and New Zealanders,” Maciej Surowiec, Microsoft’s government affairs lead for New Zealand, said in a blog post.
“It goes on to recognise that a principled approach is needed to realise this potential and mitigate the risk of unintended consequences, and as such, government agencies that sign onto the charter commit to following its principles when using algorithms to help serve the people of New Zealand,” he added.
Surowiec said he welcomed the charter’s upfront commitment to transparency and human oversight and its proportionate, risk-based approach to the deployment and use of algorithms by government agencies.
This comes as little surprise, given that Microsoft actually provided official comments on an earlier draft of the charter and now believes that the principles outlined in the finalised charter are of “fundamental importance”.
“In our comments on the earlier draft,” Surowiec said, “we brought to the New Zealand government’s attention that there has been progress in developing tools to understand and manage some of the socio-technical challenges associated with responsible AI, including detecting uses that might result in unfairness, traceability through model management, AI security, and improving intelligibility.
“We recommend agencies leverage these efforts in their forthcoming work to operationalise the charter. Our Responsible AI Resource Center may be helpful in this endeavour,” he added.
However, Surowiec also suggested that more work to support the real-world application of the new initiative needed to be done now that the charter had seen the light of day.
“As a next step, we’d encourage the New Zealand government to consider developing practical implementation guidelines, including sharing examples of government projects that have piloted the principles of the charter,” Surowiec said.
“We recommend providing government agencies that are expected to apply the charter with real-life examples that illustrate the technical, organisational and policy safeguards could help them deliver on the charter’s objectives.
“Microsoft welcomes the opportunity to provide some of our learnings from operationalising our AI ethics principles as the New Zealand government may look to best practices observed in other countries or organisations,” he said.
Further, Surowiec stressed that it was valuable for agencies to become familiar with the principles and embed them into their work, especially as algorithms are used by agencies to provide useful services to New Zealand citizens.
“The charter will be even more meaningful as additional agencies sign on,” Surowiec said, adding that peer review of the effect of algorithms can be helpful in this process.
