University of Otago Researchers Criticise ACC for Computer Model Trouble

Researchers from the University of Otago have criticised the new computer model that ACC uses to profile and target clients.

The computer model uses risk predictions in order to profile and target clients as well as to help staff manage the claims that they receive. However, it has caused controversy among some, including lawyers and advocacy groups, who have been calling for ACC to divulge details on how the program works. ACC issued a short press release, stating that the program makes three types of predictions:

  • Which clients are likely to need help and should be called proactively
  • Which type of case owner should assist the client
  • How long they expect a claim to be managed.

This description is vague and there is little detail regarding how the model works or how it is used. Some experts remain concerned that the model is predominantly focused on getting clients off ACC rather than actually helping them.

The final decisions regarding clients still seem to be under human control; a case owner manages each case. ACC has also stressed that personal details are kept private from staff.

In the modern day world companies and people are often guided by advice generated through an automatic machine. University of Otago Faculty of Law Professor Colin Gavaghan likened ACC’s system to Google’s navigation system or choosing a book based on Amazon’s recommendations; except this time the system is guiding agencies who make decisions with serious consequences for the people involved.

The main risk of the program is the potential for discrimination of clients. This possibility needs to be explored, to make sure everyone who applies for compensation is treated fairly.

The researchers did mention that there is a benefit to having agencies that are well-informed by statistics. However, while the ACC tool maybe accurate, it does not have the power to reason as humans do.

“Predictive analytics technologies show great potential in informing public decision-making, but it is important for these technologies to be evaluated and scrutinised when used in the public domain … we are calling for ACC to provide a public account of how it uses its predictive tool, so as to maintain the integrity of its decision-making,” says Gavaghan.

This article first appeared in Issue 25, 2017.
Posted 9:42am Saturday 30th September 2017 by Zahra Shahtahmasebi.