At best, our aged care overlords have got the optics all wrong. At worst, they’ve learned nothing from the worst use of AI since Skynet.
At a conference earlier this week someone asked me what I had against Mark Butler.
My answer was: “Actually, I want him to succeed so badly it hurts my heart.”
It’s true, readers. It really is.
Which is why you’ll understand that I am taking a handful of Panadol as we speak, because frankly, the latest out of the reformed aged care system is making me question my voting choices.
Yesterday, we published a story about the use of the Integrated Assessment Tool, an AI-powered algorithm that was developed, presumably, because someone deep in the DoHDA decided it would help reduce the waiting times between aged care assessments and package allocation.
As you would expect with a new technology, embedded in a new system, there have been what an optimist would call teething problems, with patients with clear care needs being ruled by the little black box as ineligible for care. Others have received IAT assessments allocating them lower levels of care than a previous assessment by a human.
Now, teething problems are one thing. But when it transpires that some time between May and November, the Department decided that human assessors could not over-ride egregious assessments churned out by the IAT, except in very narrow circumstances, that is a whole other ball of wax.
I would love to tell you that given almost 24 hours’ notice, the DoHDA came back to me with a sane, rational explanation of how and why this decision was made.
But, never mind sane and rational, to date I’ve had no response at all. And that’s despite getting a response from the Department on another matter. They know I exist, folks.
What didn’t make yesterday’s yarn was the effect this is having on the assessors who are trying to do their best by the people they are there to help.
Lynda Henderson is one of those people. She helped develop the IAT as the representative of the Older Persons Advocacy Network. Here’s what she had to say on LinkedIn a couple of days ago.
“I was proud of having been involved in the development of the Integrated Assessment Tool (IAT) for Home Care needs assessment, as the OPAN rep on the then confidential Assessment Working Group.
“It was incredibly hard work and, although I think it could be improved by a stronger focus on experiences of trauma, it’s overall a smart, flexible and useful design, in my opinion.
“People only need to go as far into it as their needs require, and it has options for further assessment to accommodate changed needs.
“The work was led by HealthConsult, commissioned by the Department.
Related
“There was no mention at the time that an algorithm would be used to classify results into the eight categories of the new Support At Home program, but we suspected as much and so built in the capacity for clinical assessors to override the algorithm if need be.
“Now we have discovered (to our horror) that guidelines have just been issued to assessors, informing them that they may no longer override the algorithm results. This will be challenged.
“To my knowledge there has been no analysis of the algorithm used, nor any details about it. So much for transparency!
“‘Robo Home Care’ anyone?”
What the hell are we doing, people?
At best, this is sloppy rules writing – rushed, ill-considered and producing unintended consequences. At worst, it is prioritising cost savings over the rights of older Australians to the care they need. A pretty shabby move for a system that has been touted as “rights-based”, no?
At best, it is crappy optics. “Let’s put the AI in charge, guys. It will save us time and money, we can spend less on human assessors, and churn through that pile of cases on the waiting list.”
Because that worked out so well when the Morrison government tried it with Robodebt.
At worst, it could cost lives, just as Robodebt did. Arguably, it was substantially responsible for bringing down the Morrison government.
Can we get serious, please? Can we put the humans back in charge, so people can get appropriate care when they need it? Can we get some robust, independent validation of the IAT with clear and transparent publication of the results?
Come on.
I’m off to see Wicked: For Good, in the vague hope of being uplifted by something this week. Have a good weekend all.