Worth Knowing

What happens if AI transforms work faster than Australia can adapt?

In the first episode of Worth Knowing, Matt and Elliot explore what AI job disruption could mean for Australian workers, and why the unemployment rate may not tell us the full story.

The Caldera Institute’s AI Work Cliff modelling stress-tests a scenario in which 20 per cent of Australia’s AI-exposed workforce experiences job loss, reduced hours or income downgrading. The result: 1.08 million workers affected and $48.9 billion in household income at risk each year.

Most may remain “technically employed”, even as their incomes fall. Australia could enter a hollow boom: productivity, investment and GDP keep rising while the wage base supporting households quietly erodes underneath them.

AI will create opportunities as well as disruption, but the warning signals are lighting up the board. What should workers do now? And how do we make the transition planned and intentional?

Chapters ➡

00:00 Highlight Teaser
00:30 Introduction
00:45 The Impact of AI on Jobs
11:21 Understanding AI Job Disruption
17:22 Industry's Role in AI Transition
19:57 Navigating the Future of Work
23:14 The Upside of Disruption
24:37 Wrap Up

Watch and subscribe to the channel now as well as our substack and on your favourite podcasting channels at worthknowingpod.com

What is Worth Knowing?

A weekly podcast about the forces shaping politics, power and technology, and how they affect you.

Matt Kendall (00:00)
And we've got a term for that in the paper and it's called the hollow boom.

where productivity and investment and GDP will still be showing at a top line level that it's really high and the the economy's looking healthy.

But what's happening is that the share of that wealth and the gains in terms of income and well being and wealth of workers is slowly being eroded away, leaving a hollowed out economy.

Hello and welcome to the very first episode of the Worth Knowing Podcast. It's a weekly podcast about the forces shaping politics, power and technology and how they affect you. I'm Matt Kendall.

Elliot Costello (00:41)
And I'm Elliot Costello.

Matt Kendall (00:43)
So what are we delving into today, Elliot?

Elliot Costello (00:45)
Yeah,

well straight to the deep end, Matt. You've chosen to start us off with how our transition to an AI-driven economy will affect jobs for humans. something that we're both quite interested in, and I know you've spent a lot of time researching and understanding. So yeah, super excited. That's gonna be the starting point today.

Matt Kendall (01:08)
Yep, it's a it's a big curly one, that's for sure. So let's get into it.

Elliot Costello (01:12)
Yeah, great. Well, I might frame up some of the politics of today and why this conversation is relevant, not just for people in our home country, Australia, but right around the world. And that's obvious on a global level. There's barely a daily conversation, certainly in the traditional news cycle and online, about the...

latest edition of Claude a new scandal or contract signed at OpenAI, all the innovation coming out of China. Of course, Elon's interview with The Economist this week. But from a sort of national perspective, it's gone from a really cold, distant issue to quite a hot one. And there's a lot of people calling for a bit of leadership from the government to not be asleep at the wheel. And we saw the awakening.

earlier this month with the announcement from our Prime Minister Anthony Albanese about protections on data and IP as we transition to a focus on AI. Effectively, Albanese has registered an AI office that sits within the PMO, the Prime Minister's office, talking about seeing the issue from both sides. I did laugh at the quote from our opposition leader, Angus Taylor, that said,

the Prime Minister talks about creating jobs and he's just created jobs within his office. I thought that was quite a good response by Taylor. But really, really what this policy does, it looks at sort of data centers and mandatory rules and controls on the data centers, so significant investment and what type of protection we're gonna put on the data centers. It looks at IP and how we protect local content and have mandatory rules.

for protection of Australian driven content, both great things. But the big sort of elephant in the room was around the mandatory rules or protections for jobs. Now that was the missing piece in this conversation and there is no rules right now and we're sort of flying blind. So I think that's the reason why we wanna commence this conversation, how AI is gonna affect the workplace and certainly.

Kick it back to you, Matt because you've done a lot of research and thinking about the implications of this transition on jobs.

Matt Kendall (03:28)
Yeah, it was you know, broad strokes, I was I was pretty happy. I think a lot of people were happy with the PM speech. I think the the the general consensus was it was great on vibes, but we

Elliot Costello (03:40)
You

Matt Kendall (03:41)
need to see the action. And I think there was fairly decent talk of action, like you said, around the data centers and some mandatory rules and IP protection and mandatory rules around that.

He did mention protecting jobs and ensuring good jobs for Australians in his speech, but there's no mention of mandatory rules. There's no mention of a strategy behind that. And I would say that the probably the reason behind that is it's it's a big it's almost like these kind of dark clouds on the horizon. It's a big amorphous thing that people don't really know what it's going to look like, n or how it's going to transpire.

And that's essentially where I where I was I was at a few months ago. and have been working on some modeling on this with the Caldera Institute to start to put some shape to that. in terms of AI affecting jobs, what would that look like in terms of wages and income, taxes?

how it's going to play out in the economy. So we did some in-depth modeling on that, where we looked into AI exposed occupations at an occupation level. starting from the point of, what the likes of Sam Altman and Dario Armodei, the the CEOs of of OpenAI and and Anthropic, respectively, the the frontier models have been saying that these

knowledge worker jobs are going to start to fairly quickly get replaced at rates of 20-30. Some have even mentioned 50%. So what we did with the modeling is we then looked at what would happen if AI exposed jobs were affected if let's say 20% of those jobs were affected. which is you know, some might say

conservative, some might say that that that's that that's not necessarily going to happen. But there's enough of a risk there based on what these these a AI leaders are saying for us to to have looked into that. And what we found out was that we're looking at around one point zero eight million workers being affected. And that transpires to being forty eight point nine billion in household income being at r at risk each year.

Elliot Costello (05:52)
in Australia.

Matt Kendall (05:53)
in Australia. Yeah.

And so if we look at the average income before the impact being about $88,000, the average income loss you know would could end up being around about $45,000 a year.

if the average income per person is $88,000, you know, that the if you're in a an an AI exposed occupation,

and you are impacted by AI, that it's it's netting out to be a potential forty five thousand dollar loss, like across the board. So some people are gonna w will be affected more than that, might lose their job and lose everything. Some people might just experience wage compaction of of a much smaller rate. But if you're sort of looking at that spread, it's that's looking at around three thousand seven hundred and fifty dollars a month and that's a mortgage payment.

Elliot Costello (06:38)
Well, what defines an AI affected occupation then? Because there's a lot of sort of hypotheticals. All models require assumptions. So just walk me through out of interest, like what is an AI affected occupation?

Matt Kendall (06:52)
we lent on the the government's own definition around this from the jobs advisory board. and it's a it's not a binary, it's a scale. So you've got, you know, a a plumber and a a copywriter is on that same scale. They're just at opposite ends of it. And then it's given a score in terms of how exposed that job is.

to AI job disruption. So what we're then looking at is then, you know, if this then starts to ripple through and you are in an AI exposed

role, that was then how we then modeled that effect of job loss or job compaction or wage loss or even disappearing entry level jobs affecting

Those workers and then the flow on ripple effects into the economy o off the back of that.

Elliot Costello (07:41)
I listen and can't help thinking you have transitioned into an AI doomer. It was probably less than three or four months ago when you explained the AI spectrum to me. In some respects, sort of sitting in the middle, I might have gone further towards the optimism under some my own assumptions now that the AI market might becoming a little bit overinflated.

Are you talking now of being a bit of an AI doomer?

Matt Kendall (08:08)
I suppose that there's two parts to that question. and I th I suppose I should explain the the AI spectrum. and the first one being, you know, there's at either end of the spectrum, you've got the AI doomers and the AI accelerationists. interestingly, most AI creators, engineers and experts and founders exist on either end of that spectrum. the AI accelerationists are, you know.

Full pedal to the metal, let's accelerate to the singularity and become one with the AI and upload our consciousness into silicon. I mean, that's that's a bit of hyperbole. But they're like feet they're like, you know, f take all the reg no regulation, let's just accelerate as fast as possible. On the other end of that spectrum are the doomers. and I it's funny because I I have few of my friends calling me a doomer, and I and I'm I'm certainly not a doomer because I would say that the doomers are very much like if we do not

regulate this now and prevent us getting towards artificial general intelligence where it becomes self-improving and self-creating. A lot of the doomers are saying that that's basically the end of civilization. That's that's the end of humanity, that essentially

Elliot Costello (09:17)
Hmm.

Matt Kendall (09:18)
it's going to kill us. So that's the that's the Doomer perspective. Then you go one in from each of those. One in from the accelerationists are the AI optimists. And they're kind of like AI

is going to be incredible. It's going to cure cancer and climate change. We probably need to regulate a little bit to you know to manage the worst excesses of it. Sometimes I'm I'm there. It sounds like you're a bit over that way now, Elliot. One in from the Doomers are the AI pessimists. And I think you can almost sort of split them into two camps. There's the what are disparagingly known as the as the Luddites who are sort of say like this technology is making

being human a lot worse. Let's just go out, touch grass, till the fields, and you know, get back to an analogue lifestyle. Then there's more the leftist perspective, which is ultimately we're heading towards techno feudalism where all the spoils are going to go towards a couple of tech companies and their founders, and we're all going to become techno serfs again.

And then kind of in the middle, and I wouldn't I I think to in the middle is is probably the wrong way of thinking about what a a an AI scout is. Because an AI Scout is this technology could be can be incredible and could achieve incredible things, but it's also

The downside risk is in is also incredible. And if we don't act now, then that downside risk starts to overtake any benefits that we could achieve with this technology. So I would say I'm being an AI scout on this. I'm kind of saying this is the and this is a downside risk. and this is the the focus of the paper is on that because

There's all sorts of incredible things that you can do with AI. I'm working with it on a daily basis. but ultimately I think we do need to as a society, as businesses, as pol our our politicians, as communities, we need to deal with the downside risks and address them accordingly.

Elliot Costello (11:21)
So,

unemployment as it stands today is not a significant problem. the rates are about 4.5%, which is statistically quite...

average. think, you know, labour would really celebrate those type of stats. What you're talking about a world in which over what timeframe we could get to 20 % levels of unemployment. But there's a whole bunch of assumptions that the market doesn't change. Like that assumption of we lose a bunch of jobs that brings down household income, up to $3,000 a week, the significant impacts, sorry,

$3,000

Matt Kendall (12:02)
So thousand dollars month. Yeah.

Elliot Costello (12:03)
a month,

Matt Kendall (12:04)
Yeah, just I suppose just also to clarify on that we're not talking about that twenty percent being an unemployment rate. and I think this is a really probably one of the most important things if if I could get across around this modeling and what what we're setting out for the government to address and understand is that around sixty seven percent of the affected workers would not appear in the unemployment statistics.

Because what we're talking about is wage compression. where you might be working five days a week and then you you know you get pulled down to two days a week. and if you are then going out there and starting your own work, you know, there's a lot of people have never, you know, have have earned salaries their whole lives. You know, it's it's gonna be a huge adjustment to then go into

entrepreneurship and actually sort of learn how to running a business. And I think there's a a l also a lot of space for for training, the government to be doing some training around that. But

Elliot Costello (12:55)
So you're talking

about government. Can I ask a question on industry? What responsibility? Look, the only reason I ask this is we.

often as constituents and voters turn to the government say it's the government's fault. Okay, I'm not defending any government right now, but what role does industry play in this? Because government can set the guardrails, it can implement good regulation. It's usually obsolete in the world of a fast moving technology, certainly in a sort of AI revolution that we're experiencing. But industry is responsible too. I learned last week in an

an Uber going to the airport that Uber offers you a job if you're a driver, contract of course, never employees, it offers you a job. have 20 seconds whilst on a freeway or a traffic light to accept the job or not and the rate. And the guy, the Uber driver was telling me if it's a low fee you can reject it.

Not just based on where it's going. It's not going in the right direction I always thought it was logistical or geographical But if it's a low fee you can reject it if you accept that low fee It will the algorithm will pull them pull the next fee down lower for you and again and again and again So kind of test you and I was staggered to hear this from the uber driver It tests you on what your threshold is for how far you willing to drive the cheaper price because uber is incentivized to give cheaper drives to its customers not to

incentivize its drivers. Is that a fair connection to industry actually playing a role in this demonetizing of income?

Matt Kendall (14:30)
I mean, I I think industry is going to play an incredible role in in this in terms of in terms of that impact. And I think it's down to the perverse incentives of the current system that we work within in terms of shareholder value. Every CEO is is tasked with the fiduciary duty of reducing costs as much as possible and increasing profits. and what AI is going to do is provide them with the means to

reduce their their labor costs. funny you mentioned the Uber example, Cory Doctorow, who is the he coined the term enshittification and wrote the book on it, which is basically, a company will first give away, you know, functionality for free to get users on board, and then we'll then focus

on customers and in terms of companies advertising and focus on them and then will then ultimately then move that focus onto maximizing profits and then thereby reducing the experience and enshittifying that experience for those two parties on an ongoing basis once they've got lock-in. he just recently with Jon Stewart gave an incredible example about a nursing app that will essentially

analyse the the credit rating of the the the nurses on that platform and if you've got a high amount of debt, you're obviously going to be more willing to take on a much lower fee for a job and having that that data and that information AI driven to basically drive wages down. So if you if you if

Elliot Costello (16:04)
What, they've got access

to the nurses or the external credit ratings. So not their personal credit cards, but their actual, their rating score. That's quite

Matt Kendall (16:09)
Yeah, yeah, yeah, yeah. The actual sort of credit rating. So if you think about that.

Elliot Costello (16:14)
black mirror. That's quite scary.

Matt Kendall (16:17)
It's inc it's

incredibly back mirror. And this is once again, if we do not have regulations around this kind of thing, this is the the the perverse incentives of companies to just, you know, maximize profits and reduce labor costs is we're basically going to get squeezed down and down and down and down. You know, people are gonna get paid less and less and less. And we've got a term for that in the paper and it's called the hollow boom. So basically, once again, top line figures, a bit like before, where we where we were saying that the top line unemployment rate might show no change.

Meanwhile, people's wages are getting lower and their incomes are getting lower and their reduced spending is starting to ripple out through the economy. The same thing is going happen at that economy level where productivity and investment and GDP will still be showing at a top line level that it's really high and the the economy's looking healthy.

But what's happening is that the share of that wealth and the gains in terms of income from and and and and well being and wealth of workers is slowly being eroded away, leaving a hollowed out economy.

Elliot Costello (17:22)
Yeah,

full disclosure Matt, I read your brief yesterday, it found it a fascinating topic. My first feeling was...

not doomsday, but like a little bit alarmist. And then interestingly, some things happen for a reason. Chris, guy, an English guy that I played tennis with on Thursday nights, I just asked how work's going. He was an engineer in the big build here in Victoria, a major sort of infrastructure project. We can do a separate episode on the big build, but I'll park that for now. But he's now moved out to Amazon investing $20 billion.

into automation and robotics in Craigieburn just north of Melbourne, an hour north of Melbourne.

And I asked him what for and he said, literally to take workers out of the workforce. that automation and robotics and $20 billion, like try competing with that. So it kind of brought me back a little bit from, know, where my first sentiment of reading your brief to like, okay, well, this isn't hypothetical what the government needs to be doing in 10 years time. It's quite real now if these, the big five or any major

players that not just building AI but deploying AI are investing that level of money for levels of automation. So I think that's my peace offering to sort of circle back and say there is a bit of a real life reality in some of the research being brought to market right now by large international and national companies.

Matt Kendall (18:54)
Yeah, well, similar similar with me of a personal story for a friend of mine who was ex Google, really smart guy. He was working for Square. And he was building their AI out for a very specific sort of product area that they were doing. and it became apparent very quickly that he was basically building AI to not just replace his role, but that basically that whole section

through the whole entire company and then not long after there was those those massive layoffs. And so there's a direction of travel here. We can all see it. That the the warning signals are there.

Elliot Costello (19:30)
It is maybe a nice way to wrap up because we promised to build a podcast that speaks to everyday people in a conversation style that's closer to a pub than a podcast studio. So I just want to take it down from the macro and maybe make it a bit more relatable. Like you've done the research Matt, credit to you. You've picked the topic you're passionate about, gone deep on it and picking up signals.

I

think that's one of your gifts is understanding trends and markets. This is a signal and a direction of travel that you're concerned about. What does it mean to the ordinary person listening to this? Like what is the message for them besides awareness around a potential transition to a new economy?

Matt Kendall (20:16)
Yeah, look, I it's I I can't like I've gone deep on the research on this, on this particular issue. And obviously I sort of live and live and breathe it through my day-to-day work as well and just watching the the narrative flows around this topic. so I I certainly can't predict what's going to happen. you know, one one of the things people say is that, you know, the new jobs will be created and they always have. and I I I would caution around that

when the factories closed down in America, those jobs weren't created overnight. in the the Industrial Revolution back in the the late seventeen hundreds, the eighteen hundreds, it took a long time for that that that technology to spread out and for for essentially sta working standards to improve for workers working in that. So I think

This is you can you can look to to history around how, that transition from a new technology killing a job to a whole new job being created can be quite extended and also quite traumatic for the people involved. And what's what's concerning about this is that with this functionality and the and the the global nature of it is that you know essentially a fli a switch can be flicked on overnight and

whole jobs could disappear, potentially. and whole industries could

Elliot Costello (21:36)
But like what

what's the message then like think you know what do you want them to take away? By virtue of listening to this

Matt Kendall (21:43)
Well,

this is I mean, this is what you know, w if you think about that, I I think we if you know, if you are exposed, if you if you are in a a line of work that is ex that is AI exposed, then you need to start thinking about a plan B, a plan C. and once again I I don't wanna move into the realm of advice because nobody really knows how this is gonna turn out.

anybody who does show any certainty on this is probably doesn't know what they're talking about. But I think, Elon Musk recently said that essentially he can see within five years AI being either as good or better at most things than humans, he had the he said apart from being humans, except for being humans. Yeah. And so I think there's a clue there is a clue in that, you know, I think,

Elliot Costello (22:24)
He did say except for being humans, except for being humans. So I watched that interview. Yes.

Matt Kendall (22:32)
you know, obviously the jobs that have

care involved are very relational or are about relationships and about one-on-one are going to be important. I think there's a still a delay on those slight physical jobs such as trades and plumbing and and craftsmanship. as much as the it can create a video, I don't think it's going to be replacing art anytime soon. but then also I think, get across being an entrepreneur,

if you're like me was in a salary for most of my life, it's really challenging to go out there and start a business. So understand what it's what's involved in in starting a business and and becoming either a sole trader at least. yeah, it's it's that's a that's a really big curly one. I think the government is

Elliot Costello (23:10)
It is, maybe it.

Matt Kendall (23:12)
gonna have to help with that as well.

Elliot Costello (23:14)
Yeah, could do a whole new episode on, you know, ideas and opportunities. The irony of part of the great replacement with AI is that it's also an enabler too. You know, my brother's just launched a new little business and I've never seen him build a website in his life and he's used AI to build it. there's, you know, opportunities to, as you mentioned around pushing yourself into new areas and taking.

the new tools available to build something you're passionate about. Awesome

Matt Kendall (23:44)
Yeah, no, I I was just gonna say I think there there is that upside and and the and look the the that upside was not covered in the paper. That was another question you asked earlier. We we just modeled that downside risk in terms of AI exposed jobs. But obviously there is a incredible opportunity for what we can do with this technology as well. And

We need to start thinking about the society that we want. That's part of that scout thing again. Because obviously our systems, our institutions are not clearly not up to adjusting to this quickly. So what do we want? What world do we want to live in? And how do we start how do we manage an intentional transition? Because if we don't start planning for an intentional transition now, then ultimately we we risk

the fabric of our d democracy, our institutions and our economy fraying.

Elliot Costello (24:38)
That's a great note to end on, Matt. What an awesome first discussion for our first episode. An intentional transition, so a great call to action for us all as individuals, as a collective within our companies, within our personal life, but also constituents and voters who hold our...

representatives to account. that was great fun, Matt. Appreciate your time and congrats on the fantastic research with the team at the Institute. It's something I'm sure others would love to read more about.

Matt Kendall (25:15)
Thanks, mate. Yeah, I I could have spoken for three hours on this, but we we have set out to keep things quite short for everyone and Each week a new episode will be released, so make sure you subscribe and and spread the word and we'll see you soon.

Elliot Costello (25:30)
See ya.