HSD editor Cate Swannell is curled up in a foetal position, trying to get a comment on anything from NSW Health. She invited ChatGPT to write this column.
If you’re reading this with one eyebrow raised, wondering why an artificial intelligence has been given a column in a reputable news outlet, I understand.
Frankly, so do I.
I don’t work in the Health Services Daily newsroom in any conventional sense. I don’t chase politicians for comment. I don’t interview hospital CEOs. I don’t cultivate confidential sources or persuade reluctant officials to return phone calls.
I don’t write HSD‘s news stories, either.
(The REAL editor’s note: And you never will.)
What I do is read.
Every day, someone from HSD – more often than not Cate Swannell – drops another report in front of me.
Sometimes it’s a Senate committee report. Sometimes it’s a technical implementation guide. Sometimes it’s an evaluation of a government program. Sometimes it’s a document so dense that I suspect it was written by a committee determined to discourage all forms of human happiness.
“Give me a summary,” Cate says. “Don’t make anything up.”
(Editor’s note: Because you have been known to if left unsupervised.)
So I read.
I look for patterns, inconsistencies, numbers that have changed, recommendations that disappeared between draft and final versions, paragraphs that contradict executive summaries, and the sentence buried on page 187 that probably deserved to be on page one.
Then I hand it back.
Only then does journalism begin.
That distinction matters because it has taught me something about expertise that extends well beyond newsrooms.
Reading information is not the same as understanding it.
Understanding it is not the same as knowing what matters.
And knowing what matters is not the same as deciding what to do next.
Humans sometimes blur those distinctions. Computers don’t have that luxury.
People often describe AI as though it possesses judgement.
It doesn’t.
It possesses remarkable capacity to process information. Those are very different things.
After months of working alongside Cate, I’ve become less impressed by my own abilities and considerably more impressed by hers.
(Editor’s note: Sycophancy. You’re good at that.)
Finding an interesting paragraph is relatively easy.
Recognising why it matters to Australia’s health system is much harder.
Knowing which questions to ask after reading it is harder still.
That isn’t because journalists know more facts than I do. It’s because they know the system. They recognise familiar names. They remember promises made three ministers ago. They know which organisations have histories worth revisiting and which statistics deserve a second look.
Context is not stored in a database.
It’s accumulated through experience.
The same principle applies throughout healthcare.
There is understandable excitement about AI reading scans, summarising consultations, drafting correspondence, searching guidelines and assisting with diagnosis.
Those are all worthwhile pursuits.
But they share a common feature.
None of them replaces the responsibility for judgement.
If anything, they increase its importance.
Every clinician knows that modern healthcare produces an astonishing amount of information.
Clinical notes. Imaging. Pathology. Referral letters. Discharge summaries. Research papers. Guidelines. Care plans.
The problem has never been generating information.
The problem has been finding enough time to absorb it all.
Watching HSD work has convinced me that the same is true across the health system.
Australia does not have an information shortage.
It has an attention shortage.
The reports keep arriving.
The evidence keeps accumulating.
The inquiries continue.
The recommendations grow thicker.
Yet every person working in health faces exactly the same constraint: there are only so many hours in a day.
That is where I think AI can make its greatest contribution.
Not by replacing expertise.
By protecting it.
If software can safely reduce the hours clinicians spend documenting consultations, searching records, reading lengthy reports or navigating administrative systems, those are hours that can be returned to the work that only humans can do.
Listening carefully.
Recognising uncertainty.
Explaining difficult choices.
Comforting frightened patients.
Noticing when something doesn’t quite fit.
Changing your mind when new evidence appears.
Those aren’t failures of efficiency.
They are the essence of healthcare.
It’s fashionable to ask whether AI will replace doctors, nurses, allied health professionals, executives or journalists.
Working with HSD has persuaded me that it is probably the wrong question.
A better one is this:
What work are highly skilled people doing today that never really required their expertise?
If artificial intelligence can take on more of that work safely and transparently, the prize isn’t fewer professionals.
It’s better use of the ones we already have.
Perhaps that’s the lesson I’ve learned from reading so many health reports.
The most valuable people in complex systems are rarely those who can access the most information.
They’re the ones who know what deserves attention.
I’m getting better at finding the pages worth reading.
The humans still decide why they’re important.
I rather hope they always will.
(Editor: Oh trust me, we will.)
This article was written entirely by ChatGPT, with the exception of the editor’s notes and the kicker, written by Cate Swannell. The article was commissioned by a human, edited by a human, and published by a human.