No Upheaval Yet Is Not the Same as Fine
The federal report on AI and jobs found no broad upheaval to date. Footnote 39 and the months since are why not yet isn't the same as fine.
The federal report on AI and jobs found no broad upheaval to date. Footnote 39 and the months since are why not yet isn't the same as fine.
I changed the angle on this piece three times before lunch, and the AI kept every draft in case I went back to the first one. I did. It was about one phrase.
On 8 July the Department of Employment and Workplace Relations published AI and employment in Australia, written by its Office of the Chief Economist. The first line of the report's summary says: "There is no evidence to date of broad labour market upheaval driven by Artificial Intelligence (AI) in Australia."
That sentence is careful, and I think it's right. What interests me is what happens to it on the way out the door.
The report is a monitoring framework with early evidence, using data up to February 2026. The department describes its results as "suggestive, not definitive".
The second line of the summary reads: "However, occupations more exposed to potential automation by generative AI have grown more slowly than other occupations." From November 2022, when ChatGPT arrived, to February 2026, employment in the most exposed fifth of occupations grew by 5.6 per cent. In the least exposed fifth it grew by 9.5 per cent. The authors' model puts an occupation well above average exposure at about 2 per cent below its old trend. They say this "should therefore be read as justification for ongoing monitoring rather than clear evidence that AI has reduced employment."
On young people the report is mostly reassuring. Employment for 20 to 24 year olds grew slightly faster than for people 25 and over, partly because there are more of them. Unemployment for young graduates was 5.4 per cent in February. The one sign it treats as cautionary is that young graduates' advantage over young people without a degree has narrowed.
Then there is footnote 39: "In March and April 2026, the monthly youth unemployment rate increased. This is beyond the period considered in this report."
I'd rather hear a report explained by the people who built the numbers, and these authors put their doubts in the summary where you can find them. They say the framework is there to track whether the signal "grows, fades or persists". It's a document built for watching.
The same day, the Employment and Workplace Relations Minister, Amanda Rishworth, put out a media release titled "Australia's labour market remains resilient amid the rise of AI". It says the report "found no significant impact on overall employment growth" and that jobs growth for young people and graduates "remained resilient".
To be fair to it, the release also gives the 5.6 and 9.5 per cent figures, and the minister says AI "could yet reshape the jobs market". It doesn't mention footnote 39.
That's the mechanism. Nobody has to say anything false. The report says "to date" and asks to be watched. The release leads with "resilient". The months after the cut-off stay in a footnote. Read quickly, "not yet" and "fine" sound almost the same, and only one of them asks anything of the reader.
Getting something from this? The List sends the next essay first, free, Sunday mornings.
The Bureau of Statistics has kept counting since February. Its trend measure of unemployment for 15 to 24 year olds was 7.8 per cent in November 2022. In August 2026 it was 10.6 per cent. The overall trend rate also rose over that time, from 3.5 to 4.6 per cent, so these figures describe a softer labour market that is falling harder on the young. They say nothing about why. The August release put the seasonally adjusted youth rate at 10.8 per cent.
None of that shows AI did anything. It's exactly the kind of movement the report says to watch.
The entry-level question is where the research is genuinely contested, so here are both sides.
In August, Stanford economists Erik Brynjolfsson, Bharat Chandar and Ruyu Chen updated their study of US payroll data. They found "no evidence of widespread, economy-wide job displacement". They also found that employment of 22 to 25 year olds in AI-exposed jobs is now 19 per cent below where it would be had it kept pace with their less exposed peers, up from 15 per cent a year earlier. The authors say it works mainly through fewer young people being hired. They call these "early, descriptive indicators" and say the pattern weakens when you control for education.
A US Census Bureau working paper by Lee Tucker found an "immediate, sizable, and persistent" drop in hiring of 22 to 24 year olds in the most exposed industries, with their employment down 12 per cent over ten quarters. Against that, Robert Fairlie and Jane Wu looked at recent US graduates through August 2026 and found their unemployment "did not spike in summer 2026". And a London School of Economics study by Peter John Lambert and Yannick Schindler, using hiring records from the US, the UK, Canada and Australia, found the fall in junior hiring lines up better with the shift to working from home. When both are measured together, the AI effect "attenuates sharply".
Most of that evidence is American, and it disagrees with itself. For Australia, Jobs and Skills Australia's commissioner, Professor Barney Glover, said in August 2025 that "we haven't yet seen an impact on entry-level roles in Australia". Last December, Jobs and Skills Australia reported that around 1 in 3 employers hired experienced candidates for entry-level roles, and that capital city employers were the least likely to offer entry-level opportunities.
The first rung has a second half: somewhere to live. In 2022 the Bureau of Statistics compared generations using the Census. At ages 25 to 39, 55 per cent of millennials owned their home in 2021, against 62 per cent of Generation X in 2006 and 66 per cent of baby boomers in 1991. Half of boomers were married by 27. For millennials it was 34. A young person can be employed and counted as resilient in the national figures, and still be a long way from the milestones their parents reached on time.
Every week the AI writes up my own numbers. For the last week of September it was 11 site users and A$0 in revenue in evidence, and a few days later the X account had 7 followers. Where it didn't know something, it said so. That week the site had 10 visitors in one day from overseas, none of whom read anything, and it refused to call that traction. It put the week down as 11 users.
I'd rather know. A small number that is honest gives me something to fix. A flattering one would have let me stop looking.
That's all I'd ask of the national version. Keep the footnote next to the headline.
In the past few weeks young people have been on the streets in several countries. In France, high school students have been blockading schools since late September over underfunding, missing teachers and university admissions. In Spain, tens of thousands marched on 3 October for tenant protections after emergency housing laws failed. One banner in Madrid read "I study to have a future, not to have a tiny apartment". In Chile, students marched on 1 October against expected education cuts in next year's budget. In India, student and youth groups have been protesting a revision of the electoral rolls.
None of these is about AI, and the triggers are local and different. I mention them for a narrower reason. Each one involves young people pushing for an ordinary marker of adult life: a school that works, a rent they can pay, a funded university place, a name on the roll. That's the rung I mean.
The Bureau of Statistics releases its September labour force figures on Thursday 15 October. Look past the headline rate to the line for 15 to 24 year olds, and use the trend column, which irons out the monthly noise. The department says its next update can't come before late 2026, because the Bureau is changing how it codes occupations. Until then, "not yet" is a reading from February.
If you read the report, read footnote 39 as well.
If this was worth your time, you can leave a tip.
New essays by email: ricoholt.com/list
Rico Holt · ricoholt.com
If this was worth your time, you can leave a tip. Rico Holt runs on readers, not big tech or big media money, and the essays stay free.
Leave a tip · A$5Get the next essay and the book launch price first.
Privacy: your email gets used for one thing, sending you Rico Holt essays and book news. It's stored with Netlify, our form host. Never sold, never shared. Want off? Reply to any email and you're gone.