Last night’s Four Corners episode, The Waiting Game, should be required viewing for every federal MP who stood on a stage during the last election and promised no older Australian would be left behind.
Nine months into the government’s “once in a generation” Support at Home reforms, the picture painted by reporters Anne Connolly and Ninah Kopel is not one of a system being rebuilt. It’s one of a system that has learned to say no more efficiently than ever before, and has outsourced the saying-no to a piece of software nobody in government seems willing, or able, to explain.
At the centre of the investigation is the Integrated Assessment Tool (IAT) and the “classification algorithm” that now determines how much in-home care funding an older person receives, a decision clinicians can no longer override.
Health services researcher Professor Kathy Eagar, who reverse-engineered the algorithm for the program, laid out its mechanics in plain terms: applicants are scored across three “branches” (functional ability, unmet need, and so-called “compounding factors” like frailty and cognitive decline) before those scores are converted into one of eight funding levels.
The problem, according to Professor Eagar, is where those compounding factors sit in the decision tree. Cognition, frailty and other health conditions are considered so late in the process that by the time they’re factored in, their effect on the final outcome is close to nothing.
More damning still: of the more than 500 questions asked in every assessment, including over 100 free-text boxes where assessors record their clinical judgement, the algorithm reads only a fraction. The rest, in Professor Eagar’s words, is essentially ignored.
She didn’t mince her diagnosis: the algorithm is not a fixable piece of software. It needs to be scrapped entirely.
The human cost of that design flaw has a face: 96-year-old Audrey Staniland, blind, living alone in Perth, who was assessed by algorithm after her provider’s rate hikes forced her to cut back on care, and was told she’d receive exactly the same funding she’d been assessed as needing twelve years earlier.
Her daughter Jan’s appeal went unanswered; after 90 days of silence, the department simply deemed it rejected. It is difficult to watch the segment and not think of Robodebt, and Professor Eagar drew that comparison herself: in both schemes, it falls to the frail, elderly person to disprove the machine, while the machine itself answers to no one.
Then there is Sean Dorney, the former ABC Pacific correspondent living with motor neurone disease, and his wife Pauline, who by the time cameras returned in the program’s second half had been waiting eight months for a reassessment that could ease even a fraction of the physical burden she carries alone.
And there is Barbara, whose family waited nearly six months for shower rails and a ramp because the government’s separate “priority” algorithm, which decides how fast approved help actually arrives, ranked her needs as low urgency, despite advanced dementia.
Out of the more than 100,000 Australians officially waiting for funding, Four Corners revealed only 88 are classified as urgent under that priority formula. Eighty-eight.
Getting a funding level out of the algorithm is only half the ordeal. Getting the actual care is another matter entirely, and this is where the government’s own numbers start to look less like progress and more like a shell game.
There are currently 98,606 Australians waiting simply to be assessed. The government points to its median wait time for a comprehensive home assessment, now sitting at 36 days, down from 43 days in 2024-25, as evidence the system is improving.
What that figure conveniently leaves out is the wait that comes after assessment: from the moment someone applies to the moment they actually start receiving the care they’ve been approved for, the median wait blows out to ten months. In 2024-25, more than 4,800 people died waiting for the funding they had already been assessed as needing.
Then there is the sleight of hand in how the waitlist itself is counted. Four Corners revealed that after the Support at Home reforms began, the department changed its methodology, quietly removing anyone receiving interim funding from the official tally.
Counted the way it always used to be, the waitlist isn’t the roughly 100,000 the government cites. It’s more than 150,000, over 53,000 people simply written out of the official picture by redefinition rather than by actually being helped.
Confronted with this on camera, Rae’s response was to insist the waitlist “has always been calculated the way it currently is”, a claim the reporter directly and correctly disputed to his face.
If the algorithm is the government’s blunt instrument, Aged Care Minister Sam Rae is its softening agent, and his performance, both under Anne Connolly’s questioning in the program and in a chummy ABC Radio Brisbane interview that aired the same morning, is a case study in having zero emotional investment in what should be a very delicate situation.
Asked directly whether he was aware most of the information collected in the IAT’s text boxes never reaches the algorithm, Rae didn’t deny the mechanics. He simply pivoted, insisting the “classification and prioritisation mechanisms draw the data that they need”.
It’s a sentence built to sound like a rebuttal while conceding the point. Pressed again on whether it was acceptable that clinical notes go unread by the system deciding people’s funding, he retreated further into procedural fog, describing the notes as merely “critical for the broader parts” of the assessment process, a phrase that manages to say everything and nothing at once.
Asked point-blank whether the current algorithm had been trialled before going live, Rae wouldn’t answer yes or no. Instead: “it was developed over many, many years”, there were “many trials”, there was “expert advice taken all through the development process”.
When asked to name the specialist clinicians who validated the final model, he offered only that “hundreds of people” were involved and that he wasn’t going to “rattle off names”. The department, it turns out, couldn’t produce those names either, because, per the Australian Society for Geriatric Medicine, no one from the field was consulted at all.
Confronted with the fact that more than 4,800 people died in 2024-25 waiting for approved funding, Rae called it “a tragedy” in the same breath as pivoting to his government’s efforts to “drive down wait times”, grief and talking point delivered in a single, seamless sentence. Told that a thousand people had lodged appeals against the algorithm, six times the previous year’s number, he described that figure as “relatively tolerable”, a phrase that will not comfort a single family still waiting.
The radio interview is, if anything, more revealing. Asked directly by host Steve Austin whether a human sits at the top of the decision to catch the computer’s mistakes, Rae insisted “there’s no artificial intelligence being used whatsoever” and dismissed concerns as “misinformation”, a strange hill to plant a flag on, given his own department was, at the very same moment, unable to tell Senate estimates who had clinically validated the “classification and prioritisation” formula he was defending.
You cannot simultaneously claim a system is too sophisticated to explain in a sentence and too simple to be worth scrutinising.
Notice, too, what never appears in any of his answers: an admission. Not once does Rae concede the algorithm has got an outcome wrong. Every failure is reframed as “more work to be done”, every criticism met with a recitation of the government’s 40 per cent funding increase since 2022, a real figure, deployed as a shield against very specific, very human stories of people going without.
If the aged care system under the Royal Commission was rightly branded a story of neglect, what Four Corners has documented feels like something colder and more deliberate. Neglect implies looking away. This is a government that built a formula precisely so it wouldn’t have to look at all: a machine that can say no to a blind 96-year-old without a minister ever having to sign his name to the decision.
It is shrewd, it is engineered to save money, and it comes with an inbuilt scapegoat every time someone asks why. That may be the most damning thing this investigation reveals: not that the algorithm is broken, but that a broken algorithm is, for a government watching its budget, remarkably convenient.
The Greens wasted no time. Senator Penny Allman-Payne, the party’s spokesperson for older people, said the investigation was a fresh indictment of what she called Labor’s reckless aged care reforms, and reiterated the Greens’ position that the IAT can’t be patched: it needs to go. She pointed out that the same tool has been used to deny funding to people with degenerative conditions like MND and dementia, and argued the government implemented an untested system with the bare minimum of accountability, in the name of cutting costs.
Allman-Payne’s message to Labor was blunt: fund aged care properly by taxing gas companies and making billionaires pay their share, rather than clawing savings back from the oldest and most vulnerable people in the country.
Her comments land just weeks after the Senate, with Greens and crossbench support, passed a bill to restore assessors’ ability to override the algorithm, only for the government to vote it down in the lower house last week.
The public reaction since the episode aired has been swift and largely merciless. Across social media, the comparisons to Robodebt came thick and fast, with commenters describing the IAT as its spiritual successor: a system, as one viewer put it, designed to be a black box that ordinary people have no hope of understanding.
Others focused their fury squarely on the minister, questioning his suitability for the portfolio and accusing him of speaking in jargon designed to obscure rather than clarify. Aged care advocates and clinicians who appeared in the piece drew praise for the reporting itself, while several posts noted the irony of a minister denying AI was involved in a “classification algorithm” his own department can’t fully explain.
Audrey Staniland’s story in particular struck a nerve, with viewers asking, reasonably, what exactly a blind, frail 96-year-old living alone would need to demonstrate to qualify for the highest tier of support, if not her actual circumstances.
Sean and Pauline Dorney’s eight-month wait drew similar outrage, with viewers noting the cruelty of a system that promises no one will be left behind while leaving a woman in her late sixties to single-handedly lift, dress and care for her husband, night after night, because an appointment hasn’t come through.
Strip away the modelling jargon, the branches and the “compounding factors”, and what Four Corners exposed last night is a government that has automated its own cost-cutting and then hidden behind the automation when the human toll shows up on camera.
Nobody in Canberra has to look Audrey Staniland in the eye and tell her no. The algorithm does that for them. And when a minister can stand in a radio studio and insist, with a straight face, that no artificial intelligence is involved in a system built entirely to mechanise a decision a clinician used to make, you start to understand that the fluff isn’t just in his hair. It’s the whole defence.
Labor inherited a bad system and vowed to make it better. Instead, they have made it worse. The most disturbing part is that none of this looks like an unforeseen side effect. These are cruel, deliberate design choices, made by people who understood exactly what they would do to the most vulnerable, and made them anyway.
Sam Rae doesn’t appear to have any empathy. He’s robotic.
I have stated this previously: Mr Rae need to be removed from this portfolio.
The Aged Care system has not improved to the levels that it should assists the older person in getting help.
South Australia was labelled ‘ Ground Zero” in relation to Abuse and Neglect in Aged Care. Hence the two Royal Commissions, Aged Care and Disability Care. Abuse and Neglect still occurs in Aged Care with Cover Ups rife. The government’s systems are still way off to deal with these issues. The federal departments system and policies are poor and are laden with mistakes.
South Australia legislation pushed through parliament, with little consumer voice, is and will lead to worsening Elder Abuse.
I am convinced once a person reaches the age of retirement or get to 70 years of age this government doesn’t want to know or hear anything about you. As for ai that is what is running this damn country because the politicians have no idea what they are doing.
I have never seen anybody who can avoid a question like Sam Rae can. Maybe that is why he was given the job. The real problem, of course, is that people are — ever so selfishly and brazenly — getting old. If they would just stop being so reckless and unthinking, Sam would not have to work so hard on his sidestep. Poor thing, I feel sorry for him. I wonder, though, if he has realised that he, too, will one day be old.
One day Sam will, indeed be old. His ministers pension will ensure, however, that he will be able to afford the care of his choice without having to rely on the state to assess or provide. Such is the privilage of the haves Vs the have nots.
The interview simply confirmed what we already know. Following the debacle of the ‘Nurses Pay’ fiasco, it should have been clear that reliance on algorithms and IT generated figures is NEVER accurate. The human factor has not been taken into account…but then I suppose it’s just another way of de-humanising, which seems to be the flavour of the month.
The IAT can be kept as a standardised assessment tool and decision support tool but not allow it to be the final decision on a person’s care needs.
The clinical assessor should be able to override the IAT outcome where it does not reflect the person’s actual complexity, safety risks, functional limitations or reliance on unpaid care.
The assessment should consider :
– whether the person can manage safely and reliably , not simplify whether they can perform a task
– the combined effect of multiple conditions
-how many hours of unpaid care family or friends are providing
– what is likely to happen if that support is not provided.
Any clinical override should be documented, independently reviewed, and used to identify where the IAT itself needs improvement.
The principle is simple: the IAT should support clinical judgement, not replace it. Importantly, a person should not be assessed as having low care needs simply because someone else is providing substantial unpaid care to keep them safe at home.