The CSIRO has confirmed that its data branch Data61 is no longer working on the government’s controversial welfare drug testing scheme, as the first location for the trial has been revealed.
The government said on Tuesday that Canterbury-Bankstown in western Sydney would be the first of three locations where new welfare recipients are to be drug tested from the start of next year.
It was originally announced that Data61 would be closely involved with this decision-making process, developing a “data-driven profiling tool” for the Department of Social Services to zero in on suitable locations.
But a spokesperson for the CSIRO confirmed that Data61 was no longer involved in the scheme, and had provided only “general advice” on machine learning, rather than developing the algorithm.
“Data61 has been in discussions with DSS about the scope and involvement in the DSS drug testing for welfare payments trial,” the spokesperson told InnovationAus.com.
“The outcome of this discussion is that Data61 has provided general advice to DSS on the use of machine learning for possible selection methodologies to determine which people in a region should undergo drug testing trials.
“The application of this advice is at the discretion of DSS.”
Data61 is no longer involved with the politically-charged scheme, the organisation confirmed.
“Data61 has not been involved in selecting the regions that have been announced. Currently no work remains to be delivered by Data61 for this project,” the spokesperson said.
“Data61 remain available to provide further advice to DSS, if required and at their request”
The government had previously announced that Data61 would be far more closely involved with the scheme in developing the tool to select the three trial locations.
“We’ll use a combination of data that we will help develop with Data61 and CSIRO,” Social Services Minister Christian Porter said in May following the budget.
CSIRO digital executive director David Williams also outlined more involvement from Data61 during senate estimates in June.
“They asked for Data61’s support in doing the analysis to see whether predictive analytics could help them in that task,” Mr Williams said.
Information released to InnovationAus.com under a Freedom of Information request showed that Data61 was sent a formal contract to develop the “prediction system” in mid-May, but this was never signed by the organisation.
“No data analysis or models have been developed or deployed by Data61 for the purpose of this project. Therefore, no contract was executed for this, because no work has been delivered for this project,” the CSIRO spokesperson said.
The CSIRO was first contacted by the Department of Social Services at the start of April, when it was asked to “discuss building a profiling tool for DHS to use looking at certain characteristics of the recipients to choose for the trial”.
Despite no contract being signed, Data61 confirmed that it had conducted “preliminary technical analysis” of de-identified data provided by the department which helped “determine the design of the full project”.
Along with the machine learning consulting, this appears to be the extent of Data61’s involvement with the welfare drug testing scheme.
Mr Porter and Human Services Minister Alan Tudge said the area in western Sydney had been selected for three main reasons: the “real problems with drugs in the community”, the presence of support services and a “large number of people entering the welfare system”.
They said the government had utilised a “range of best available data” to reach that decision.
About 1700 new welfare recipients in the area will be drug tested from the start of next year, provided the necessary legislation passes through Parliament.
The government has now been forced to negotiate with the crossbench after Labor and the Greens withdrew their support for the controversial scheme.
The Department of Social Services was contacted for comment.