The Business of Data Festival welcomed Ian Oppermann PhD, Chief Data Scientist for the New South Wales state government in Australia, who shared insights into the founding of the country’s first government committee on AI
In March 2021, the New South Wales Government formed a committee with the key remit of advising on the appropriate use of artificial intelligence.
The New South Wales AI Committee is chaired by NSW Chief Data Scientist Ian Oppermann PhD. Its members include data leaders with expertise in law, ethics, technology and AI. A former Australian government privacy commissioner and the CTO of Microsoft in Australia are among its ranks.
As part of the inaugural Business of Data Festival held in July, Dr Oppermann, who is also a Business of Data 2020 ‘top 100’ innovator in data and analytics, sat down with us to discuss the committee’s purpose and plans.
He told the festival audience how the formation of the NSW AI Committee followed the release of the NSW’s AI strategy in 2020, which lays out plans for the state to take advantage of the AI opportunity responsibly.
“The AI strategy was actually the result of quite a lot of work leading up to identifying all of the different issues that we thought we needed to put into place in order to appropriately use artificial intelligence,” Dr Oppermann said. “We looked at ethical issues, we looked at issues related to trust, issues related to governance and assurance and realized what we needed to do was re-ground the conversation essentially into an appropriate-use-of-data conversation.”
When the strategy’s authors took the policies to experts for discussion and critique, they noticed gaps. One such gap, Dr Oppermann said, was the need for stronger engagement with standards. The feedback made Dr Oppermann and his team recognize the need for a group of people to constantly provide friendly criticism, and the AI Committee was subsequently established.
Building a Framework for AI Application
The Australian government’s thinking around the best way to capture the opportunities data, analytics and AI are creating in government is evolving as the use of these technologies in the business community matures.
Following its creation, one of the committee’s first tasks was to develop an AI assurance framework that the government can apply to all its AI-related rollouts.
“People have been developing machine learning and AI solutions for some years now,” Dr Oppermann said. “So, what we’re doing as [a] first order of business is looking at those existing projects as we develop an assurance framework. So, borrowing from the world of standards and borrowing from ICT assurance.”
“We’re looking to see how we can better inform a generic assurance framework, and then for those existing projects see what we can do to improve their quality,” he continued. “For future projects [we want to] provide people with the toolkits they need so they can appropriately design the use of the AI project, work out where to get the data from, work out which datasets they should use, understand issues around data quality [and] understand issues around the appropriate use of data-driven insights.”
The assurance framework is set to borrow from the world of standards and ICT procurement and bring those elements together with considerations around risk, harm and ethics. It will be continuously tested with teams contemplating building new AI projects.
Paving the Way for Data Sharing
Dr Oppermann expects that by applying a common framework to help the government better navigate AI projects, another important area of data and analytics maturity will be unlocked: Data sharing.
“The big issue for 2021 is data sharing and use. I mentioned that one way of seeing AI is just a use case of data sharing,” Dr Oppermann said. “Last year, we released the Smart Places strategy, [our] smart cities and smart places strategy. It’s a use case of data.
“In general, what we’re trying to do is reframe the conversation around outcomes, not just doing something but really driving outcomes in the real world. We’re building indicators to underpin those outcomes, so we can be very clear about what we’re measuring and what we’re trying to achieve.
Oppermann said the tide of data and analytics use has been growing in New South Wales, especially in the past 5-6 years, citing existing use cases in combating family and domestic violence, transport optimization and planning.
“Now, with the Smart Places strategy and with the AI strategy, we are really accelerating our use of data,” he said. “If we’re going to accelerate our use of data, we need to put in place all the frameworks and all the systems and think about the whole problem; the whole challenge:
“All the way from understanding how to maintain and build trust with the public, to data governance [or] data quality issues, to the appropriate use of AI. It’s all coming together this year.”