The NDIS Bill opens the door to another Robodebt disaster

Currently, the NDIS allows people with disability to be considered as individuals, not simply as categories. However, the Bill reduces which aspects of a person’s life can be taken into account. It then enables even discretionary decisions to be made automatically, rather than by an experienced staff member. 

This is a radical change. The Government’s review of the NDIS in 2023 found that “we must ensure the NDIS experience is centred around the whole person and their disability-related support needs”. 

The Bill will make that much more difficult. It makes effective NDIS oversight even more important than it is now. Yet the Bill also winds back many of the usual protections that apply when a human makes a decision.  

Our concern isn’t abstract. Just last week, Parliament heard damning evidence about failures in the tool used to assess aged-care funding packages for older Australians. The fact that human decision makers can’t change automated decisions has been identified as a major contributor to these problems being made worse.

Humans aren’t always better than machines at making decisions. But we need to take much greater care in targeting the decisions government makes automatically, and the safeguards needed when government restricts human oversight. 

If the Bill only allowed for the automation of simple, non-discretionary decisions, there would be less concern. For example, it’s faster and more reliable to use a calculator for a complicated calculation, than to do the task longhand. This might be what the Government means when it talks of automating consideration of “low-value, routine claims … where clearly defined criteria are met”. 

However, the Bill doesn’t only permit these sorts of decisions to be automated. It expressly allows decisions involving “discretion” and “evaluative” judgment to be automated. Automating such decisions is more dangerous, because there isn’t a single right or wrong answer. Instead, the task is to weigh up factors and reach a decision that is preferable in the circumstances.  

For example, the Bill would allow a person’s support needs and access to NDIS funding to be decided automatically. This calls for human judgment, to ensure proper consideration of the person’s individual circumstances and how their disability affects their lives.

The Bill requires the Government to introduce guidance for the new automated decision-making process. But the content of that guidance isn’t spelled out in law. It cannot be left up to the Minister or CEO of the day to decide when or how to automate decisions, without strong legal guardrails protecting human judgment and oversight. 

Currently, important NDIS decisions are subject to internal review by a more senior officer. This is necessary, because a second set of eyes can ensure all factors are weighed appropriately when deciding what support is needed for a person with disability. 

The Bill removes this regular form of internal review for automated decisions, and instead allows the CEO of the National Disability Insurance Agency to intervene when they see fit to do so. In other contexts such as for immigration visa decisions, when internal review is replaced by an override mechanism, it is used only in highly exceptional circumstances.

To its credit, when the Albanese Government first came to power, it quickly set up a Royal Commission into the Robodebt scheme. It was committed to understanding what went so terribly wrong, and to making sure these mistakes were never repeated. 

The Prime Minister himself was shocked. On the day the Royal Commission was released, he described Robodebt as “a gross betrayal and a human tragedy.” Half a million Australians were affected. It caused untold stress and anxiety, and in some cases contributed to self-harm. Addressing this injustice ultimately cost taxpayers about $2 billion. 

And so it’s baffling that the NDIS Bill would enable automated decisions in such a sensitive area - with so few safeguards. But it isn’t too late to address these problems. Fixing the Bill would involve spelling out more clearly what sorts of decision are, and aren’t safe, for automation, and introducing effective oversight and review.

Professor Edward Santow served as Australia’s Human Rights Commissioner from 2016-2021. He and Sarah Sacher work at the UTS Human Technology Institute.