NSW government looks to Microsoft AI to cut $10m costs
The NSW government is working alongside tech giant Microsoft to develop a new machine learning tool that will automatically monitor its $30 billion annual procurement budget and tell it when money is being wasted and how it could cut costs.
The technology is being developed in a partnership between Microsoft and the NSW Data Analytics Centre (DAC), with the hopes of turning it into a commercial software service which can later be sold to other governments in Australia and around the world.
"I call it colloquially the procurement sharpener," NSW Minister of Finance, Services and Property Victor Dominello told The Australian Financial Review.
"We have a procurement spend on goods and services of around $30 billion a year and if you can sharpen that and get 2-3 per cent savings, you can drive those efficiencies into other critical areas of public policy. We need to make decisions based on information, rather than gut feel."
Microsoft and the DAC have signed a non-exclusive memorandum of understanding for the commercialisation of the technology and collaboration on other data science projects. Under the deal the DAC has used Microsoft Azure and a range of Azure cognitive services to build the tool.
NSW's DAC was formed in 2015 and is led by chief executive Ian Oppermann. It works with NSW government agencies to develop insights and tools to solve policy and regulatory issues.
A proof-of-concept of the AI categorisation tool has already been developed and focuses on the 8 million transactions which make up the NSW government's procurement budget.
If the NSW government saves 2-3 per cent a year in its procurement budget, that equates to $10 million.
Microsoft Australia managing director Steven Worrall said the partnership was possible thanks to Microsoft recently achieving certification from the Australian Signals Directorate to host top security government data sets in its public cloud.
He also said NSW's status as one of the biggest governments in the region (its budget is larger than some federal governments in the Asia Pacific region) meant it would be attractive to governments around the world.
"This is a consequence of the hyper-scale cloud computing world we find ourselves in. That kicked off within the last 10 years and has now become popularised," he said.
"If we didn't have a government certified level of security that provides assurance of how the data will be used and stored and has that trust built in, we wouldn't even get out of the starting blocks [for a project like this]."
If the proof-of-concept goes well in the next six months, Mr Dominello is keen to expand the AI technology to other fields to tackle problems such as insurance fraud, builder insolvency in the construction sector and skills alignment between university and TAFE courses and the jobs being created.
"How often do we see people studying courses in TAFE and thinking if they do well there will be a job at the end of the pipeline, but then there is no work in the field. This could lead to an analytics tool ... that corrects that mismatch," he said.
"The next five years are going to be unbelievably exciting. I'm super passionate about planning and we're also developing a digital platform where you can get real-time information on movement in the city which will drive infrastructure investment in railway lines, roads and even schools and where to put them."