The Joe Walker Podcast - What Transformative AI Could Mean for Australia — Danielle Wood

Episode Date: June 3, 2026

Danielle Wood is an Australian economist and the current chair of the Australian Productivity Commission.See omnystudio.com/listener for privacy information.

Today I'm speaking with Danielle Wood, who is the current chair of Australia's Productivity Commission. Danny, welcome to the podcast. Thank you for having me, Joe. So just as quick context, how would you describe the Productivity Commission to people listening in from the UK or US? Because it's kind of internationally speaking, it's kind of a unique institution. Yeah, we are pretty unique. I mean, I describe us as the government's think tank. So, you know, really our job is to make sure that government has kind of high quality evidence back.

information in front of it and we span obviously economic policy and productivity but also social policy, environmental policy. But we look a little bit different in the sense that we're also designed to be independent. So we sometimes have to tell government hard truths or challenge them in terms of where they're at on policy. And so our independence from government allows us to do that. I remember reading a story about how Gary Banks, who was the first chair of the Productivity Commission, discovered that in a gambling reform debate in the US, people were citing evidence from the Australian Productivity Commission because all of the evidence in the US was kind of paid for by different lobby groups of vested interests, so people just didn't know who to trust.

And so they ended up citing evidence from the Australian Productivity Commission. I love that. Yeah. And I mean, that really goes to, you know, our job is to bring that national interest length. and there are just so many policy debates which are like gambling where you have these really active vested interests. But our job is to kind of pull together all the arguments, all the evidence, do that broad concentration and then give government our best version of what's in the national interest.

So today we're going to chat about AI. And there's not going to be an exhaustive chat about the economics of AI. We're just going to kind of poke around a few different topic areas that I find interesting at the moment. But generally I'm very excited to hear to learn. and how you're sort of making sense of this unfolding technology, even just in very broad, kind of gestalt ways. So first question is, I was catching out with a friend on the weekend,

and we're talking about how big of a deal AI is going to be. And I said, and when I use subjective probabilities, I'm always being semi-ironic, but I said I give AI a kind of 90 to 95% probability of being at least as big as the internet, and maybe a 20 to 40% probability of being, being at least as big as the industrial revolution, but which I mean that it could lead to a regime shift in the underlying total factor productivity growth rate. So I don't know, maybe

the growth rate moves to like three to five times the historical average in the same way that through the industrial revolution, it went from nearly zero percent to modern rates of about 1 percent or above. So 90 to 95 percent probability of being at least as big as the internet, 20 to 40% probability being at least as big as the industrial revolution. How do those probabilities sit with you? Look, I mean, I think that is the right way to think about the kind of plausible set of outcomes. I mean, we, you know, we think it will have a meaningful impact on productivity. It's, you know, general purpose technology.

Like the kind of previous waves, the internet was one, you know, ICT more generally, electricity, steam, you know, all of those things have through history fundamentally kind of transformed the economy and touched different sectors. The kind of conservative view, which is probably your internet parallel, is hey, there's just a whole lot of different tasks that it can do more efficiently. And we did this exercise for some work we did for the government last year on productivity that said, okay, well, let's look at the evidence on these sort of tasks specific efficiencies. Let's look at, you know, the number of tasks across the economy that might be affected. And you get a kind of, well, it could improve

labor productivity by about 4% over a decade. So that's like pretty meaningful. It would actually be almost a doubling of where we've been over the past decade, albeit from quite a low, a low base. The industrial revolution parallel, your three to five times total factor productivity is more of a fundamental change than that. And I think you start to get that sort of scale of impact in a world where AI actually changes the innovation process itself.

So we know that over time it's innovation that drives total factor productivity. if it makes us better at doing innovation, if it starts to speed up the rate at which we find new goods or new ways of doing things or ways of solving social problems, then you're in that kind of extraordinary world that you're talking about. And I think that there is a real chance. I'm not good at putting probabilities on things.

I don't know if I'll be as specific as you have been. But I think that is a world that we could be in with this technology. I think that's the crux, right, between the kind of optimistic and then super optimistic case. It's like whether AI can automate R&D itself. So 4% labor productivity growth over the next decade. Is that like the upper bound or the base case? No, I mean, I think we said that's a kind of conservative case. Okay.

So I was using a pretty standard methodology. It's sort of taking conservative end of estimates on task-specific efficiencies. So, you know, I think that is very achievable. So that's 0.4% per year on average. Yeah, that's right. As an increased. Yeah, and that's about what we've been growing at on average over the past decade. Yeah, so hence roughly doubling what we've done in the last decade.

And is there a more optimistic scenario there? I mean, we didn't model one. Okay. But, you know, there's certainly people around that have come up with, you know, closer to the sort of, you know, three to five times TFP estimates that you're talking about, but that does, as I say, rely on, you know, much more impactful AI in terms of changing innovation processes. So let's talk about the labour market implications of AI.

So until about a month ago, Daria Amaday had been going on TV, saying that AI is going to eliminate up to 50% of all entry-level white-collar jobs. And over the last two to three years, I've been talking to a lot of my technologist friends, some working in AI labs. And there was always this view that AI is going to bring about this very sudden job apocalypse. And I always had this, well, not always, but increasingly had this sense that the economics were going to be much more subtle than that. And my technologist friends would have benefited from talking to my economist friends,

and vice versa. But I think that has increasingly happened over the last year or two. But what are some frameworks that you like at the moment for thinking about AI's impact on jobs? Yeah, great question. And I think you're exactly right. I think the technologists, you know, they're at the cutting edge. They're living and breathing it. They may be seeing it play out in a very small set of networks where technology adoption is much faster

and looks pretty different to the rest of the economy, let alone, you know, as we start to talk about shifting preferences and things as we will in a moment. But the kind of, the earliest framework that's been around for thinking about this and the earliest kind of modelling on jobs effects was really around, you know, what happens to particular tasks. And so jobs and skills in Australia has done this exercise for Australia, which basically divides every job into task and says, okay, which tasks are the ones that are kind of most likely to be automated by AI. And then they say, well, what's the bundle for particular jobs? How at risk is a particular job given that set of tasks? So what they found when they did that exercise was they said

about 4% of jobs in Australia were highly at risk of automation, i.e. most tasks can probably be done by AI. So things like data entry, you can imagine those jobs probably disappear. And from a policy perspective, we have to, you know, work to look after those workers and transition them. You then have a bigger set of jobs, which was about 31% in the Jobs and Skills Australia estimates, which likely to be augmented. Some of the tasks can be done by AI. Some are human. And there really the challenge is kind of how do you upskill people to be working with the AI?

So that's the sort of, I think, the kind of first cut framework people use. There has been now kind of a lot more thinking about, you know, what I would call kind of second round effects and more sophisticated models. So one is really that it sort of depends on the number of tasks and how tightly those tasks are bundled together. So there's a new book coming out called Messy Jobs, which is really this idea if you've got a lot of different tasks and they're quite different, that's going to be harder to replace with AI. And the tight bundling is really about, you know, are there some of those tasks that are still very much human tasks? And they can't be separated even from the ones that can be done by AI. And, you know, the example that sometimes used here is the radiologist. You know, AI can do the scan reading better as well as probably better than a radiologist.

So everyone said, OK, well, these guys are, you know, not going to exist in five years. But what we've seen actually, at least in the US, is the numbers have gone up. And the idea is that the tasks are tightly bundled. It's the communication with patients. It's the accountability. If something goes wrong, I need to have somebody who takes responsibility for getting the answer wrong, including through the legal system. And so we still need that person.

And so they still end up kind of having that same span of controls. That's tightly bundled. The second point is around how do these things actually kind of play out in markets. So if we imagine that we've got a lot of augmentation going on, things are getting more efficient to produce, those services become cheaper, how do consumers respond to that? And in some sectors, if you've got kind of high elasticity of demand, we're actually going to consume more of that product and whatever labour is left will be in higher demand. So it's not always clear cut that just because some parts of the job are replaced by AI,

that labour demand will fall. So you've got that consumer response that you need to build in and that's going to look really different across different sectors. And then another way of thinking about it, which I think is really fascinating, a little bit more speculative is through the lens of how might this start to shift preferences in the longer term. And so in a world where we have abundance of a lot of things,

you know, we can read all of the, you know, AI-generated books for zero cost. You know, we can access any music in the world for nothing. This is all being generated by AI, it basically zero cost. Might we start to value the kind of human element and things produced by humans more? So this is a fantastic, really interesting academic. I'm trying to pronounce his name is Alex Imas.

Yeah, I think that's right. Yeah. So he's written about, you know, what we might see is the kind of growth of the relational sector. So things that are kind of uniquely human, which is everything from care to creative outputs to craft, artisan bread, you know, whatever it is, things with that human touch might actually be more in demand and more valuable, which in turn supports labour market outcomes in those sectors.

So it absolutely is more nuanced, I think, than those kind of tech doomsday scenarios. But much more interesting, much more exciting, but of course lots of implications for policy and all of that. Absolutely. So interesting. Okay, let me pick up on a few. few different threads. First one is the bundling, which I think is interesting. And you mentioned

radiologists, which is a canonical example. Another good example is, I think even a few years ago, GPT4 was scoring in the top 10% of takers of the bar exam. But being a lawyer is more than just repeatedly taking the bar exam. And there's this very tight bundle of different parts of the job. And actually, I think if one or a few of those tasks in the bundle can be automated, it makes the worker more productive and so it should actually increase wages. So another interesting implication of that is obviously employers have an incentive to automate jobs for which there's just like one task left in the bundle to automate because then they're going to immediately reprimand. all of the cost savings of that. So it seems like you'll, like for a lot of roles that are being automated, a lot of roles for which their tasks are being automated,

the wages go up and up and up until the last moment where the role then just gets fully automated. So there's this kind of like everything's looking rosy and then displacement dynamic. Does that make sense? And then is that, how should policy makers be thinking about that? Look, I think it does make sense for some roles, but not all. And I think, you know, they're just, there's still going to be a whole lot of roles for which that kind of judgment, accountability piece just cannot be automated. You know, until we, you know, have a totally different set of,

institutional settings around accountability and, you know, that's kind of a long way. We're still going to need someone to blame. Yeah, indeed, indeed. And I think, you know, we still, for trust, are going to want humans in the loop in a whole lot of areas. But there probably is a subset of roles for that kind of pattern that you've just suggested might hold true. You know, truck drivers is another one of the examples where people have been saying for a long time or these jobs will go. So, well, yes, the machine can drive the truck. You know, increasingly as robotics improve, maybe AI plus robots can, you know,

load and unload the truck. Looking after the truck, the kind of safety thing, making sure that the truck doesn't get stolen, the load doesn't get stolen. You know, those are things which people have found quite hard to replace. But maybe, you know, if that's all that the kind of human is doing that's left, maybe it is worth the investment in trying to find alternative ways to do that and you get more of the kind of scenario that you're talking about that kind of tips over into being fully automated.

So I suspect there may be certain roles and professions where that happens and it just means you kind of hit a tipping point and then you lose those roles. And so policymakers, of course, you know, as always, when there are these sort of structural shifts have to think about how do you create opportunities for those people elsewhere. But I don't think that will be a kind of across-the-board phenomenon. Yeah. So the truck drivers and warehousing personnel examples are interesting because they're towards the lower end of the skills distribution,

which is kind of a narrative violation given so much of the talk is about knowledge workers being displaced. It's interesting, I guess this is more of a comment than a question, but about a month ago, I'd record an interview with Greg Clark, the economic historian. I'll put it out in a month or so. But one of the surprising insights from his book on the Industrial Revolution A Farewell to Arms is that the biggest beneficiaries of the Industrial Revolution

were unskilled workers. So their wages rose relative to skilled workers. I think the three reasons he gave for that were, firstly, like a lot of kinds of unskilled work involved dexterity, which we haven't mechanized yet. Secondly, a lot of them require social intelligence, which again, very hard to mechanise. And then thirdly, the demographic transition began a bit before the industrial revolution and that curtailed labor supply. But, yeah, again, just kind of counterintuitive, be unskilled with the biggest beneficiaries

in terms of the relative rises of their wages. But now this transition seems that it might be different to that potentially. Yeah, I think, I mean, that's really interesting because it's sort of the opposite of the narrative that I have heard that, you know, over time we think of a lot of technologies have been skills biased in the sense that they've displaced the kind of lower paid, less skilled workers and created the job opportunities for higher skilled workers. So that's a really interesting counter narrative. I mean, I think most of the narratives I've heard on AI is, it's not clearly skills-based. It's more about the nature of the work. So if you're doing something that is kind of repetitive, if it's, as I said, rather than a kind of big bundle of task, it's sort of one task. So this is the kind of computer coders example, much easier to replace.

And you can imagine some of those jobs are at the sort of lower income, lower education end of spectrum. at the Hota is a very well-paid, highly skilled work that could be replaced. So I think we're expecting less of that kind of skills bias. I think the bias that people are more worried about is, you know, what I call the seniority bias, which is the point that you raised before. If the skills that are not easy to replace are around judgment and expertise and accountability, how are we going to train new people coming into a profession or an area where historically the way they have built those skills up is doing the grunt work.

And that grunt work is the work that is easily picked up and replaced by AI. So I think, you know, to the extent that policy makers are worried about biases in technology, it is this kind of question of seniority bias and are we going to wipe out job markets? for a whole lot of young people coming into different professions. So on the more speculative aspect of the demand side, am I right in thinking that the reason there will be, or there likely will be elasticity of demand for what Alex Imus calls the relational sector, is that many of those goods and services are inherently positional?

And so they can't be easily, like, demand for those can't be easily satiating. Yeah, it's right. So it's the scarcity in itself is the value. Yeah. So, you know, lots of really interesting studies that show, you know, I value something more if other people can't have that same thing. And so, you know, if I want to use the, you know, best personal trainer that I can see on Instagram, you know, by definition, not everyone can have a session with that person. and all trainer, you know, I can signal my taste by, you know, getting a tailored coat or consuming a particular type of beer. You know, so there's all of those very human impulses still exist in this world. And what scarce is this kind of human touch.

He also has this cute point, which is that in this world, bowel mulls-cost disease becomes a feature, not a bug, because it's a way of keeping humans employed. Exactly right. Exactly right. So yeah, it's what keeping people busy. And yes, their wages are high, but hey, we've got abundance everywhere else. So the economy supports that. So let's talk about AI and industry policy. So let me just tell you generally how I'm thinking about industry policy at the moment. So I can see good reasons for using it in principle, like if there are positive externalities or if there are coordination failures or if the government needs to provide some public goods. And it seems like the main critiques are very kind of. of practical and like heavily context dependent. And so if you can execute industry policy well, then it can be good. I think the thing I just don't know yet is whether there are any like best practices

that we can draw out of the success stories because I want to know about the graveyard of failures and whether they were also using the best practices. So at least to this point, am I thinking about it so far? Am I asking the right questions? Look, I think you are. And I mean, so what's industry policy? It's kind of trying to nudge economic activity in a certain direction or support new or existing sector. And, you know, why might we do that?

You know, you kind of pointed to the sort of traditional set of arguments. There might be market failures. It could be, you know, learning spillovers. It could be agglomeration benefits. There could be, you know, need for certain public goods. So, you know, I think those arguments are well known and well accepted. I think it is an implementation challenge, but it's very real. So it's, you know, by definition when governments are doing this,

If you're doing something more here, you're doing less of something else, which is often not thought about when we're thinking about industry policies. So we, you know, in a world of constrained capital, constrained labour, if we're pushing people to this part of the economy, this part is shrinking. So it's actually really important to think about the kind of net economic impacts. And so many of the studies of success or otherwise focus on, you know, were we successful in growing this sector? but when studies have gone and looked at the kind of global economic impacts, they said actually, overall, you can't actually see any benefit or it was negative overall. There's a fiscal cost if we're talking about, you know, some of the kind of demand, sorry, some of the kind of production subsidies or grants or any of the levers that governments might use, and we got to take into account the cost of that as well. and, you know, there are just a lot of implementation risks. And the biggest one, I think, is that you end up creating a whole set of firms that are very reliant on government intervention.

You know, they want that grant, they want that subsidy, and they waste a lot of resources going to governments and trying to find ways to get governments to do what they want. and, you know, that is money that's not spent on innovation and pushing forward. So you have real risks kind of distorting overall economic patterns, which I think need to be taken into account. So it's not to say that you never do it, but I think you do it pretty sparingly and where you think you've got a really solid case. So in addition to what you've just said, what are some lessons you think we should draw from our experience of supporting the car industry? Look, I think a really important one is the need for off-ramps. And, you know, if we think about some of the reasons we just said that you might support an industry like auto-manufacturing, it is around learning spillovers.

So it takes us a while to work out just how we do this and different businesses benefit from those learnings. we might be creating a kind of ecosystem where you have multiple different firms and supply chains and workers that shifts you down the cost curve over time. But all of those are temporary arguments for support. There's very little in the way of kind of long-term case for government support and industry policy except for, you know, a subset which is around national security resilience. But, you know, that was not the car industry arguments. And I think the challenge was we didn't. We weren't economic in a global sense at producing cars. And so governments stayed on the hook for those types of production subsidies. And increasingly they found it

really hard to withdraw them. Those companies had a lot of sway. There was obviously a lot of workers in those sectors that had to go through difficult transitions once that government support was pulled out because the sector just wasn't capable of standing on its own feet. Do you have a sense for whether AI at the kind of category level, so not thinking yet about specific industry policies, but whether AI at the category level will pose unique challenges for industry policy? Yeah, there's different levels that you can look at it. So I'm really interested in the kind of diffusion question, which is, by the way, you know, the kind of the vast bulk of innovation policy. in Australia is around diffusion, like about one to two percent of what we do is new to the world innovation. A lot of it is kind of picking up what is elsewhere and embedding it inside our

sectors of the economy. Which is kind of what you expect for a smaller economy and middle power, right? Exactly. I mean, you know, great that we do that new to the world, but, you know, so often I think we pick up policy debates from the US or other places which just look a lot different. So the diffusion question is really important. You know, should we think about industry policy for kind of a set of AI applications or, you know, areas where we might play in the AI supply chain,

I think it's hard to justify outside of, you know, what I would call kind of horizontal policies, policies that apply to new firms and innovative firms but aren't AI-specific. So, you know, government does, you know, a lot of R&D support, and that's about trying to get the benefits of innovation and spillovers. There was, you know, a number of things in the federal budget this year, actually, around loss write-offs for startups, which is about kind of cash flow for new firms. there were things around early stage of virtual capital incentives,

those kind of things that say, okay, well, we know that new fast-growing firms are a good thing in the economy. We're going to try and address some of the pain points for those types of businesses. A lot of those, you know, AI firms will fall into that category, but it's not specifically targeting AI firms or AI applications, if that makes sense. That makes sense. And, you know, and that is consistent, I think, with a lot of the literature on industry policy that suggests doing those kind of horizontal

type policies that are good for business in general or good for innovative business, will tend to have bigger benefits and less risks than doing really targeted kind of picking winners interventions. It strikes me another way that AI might pose unique challenges for industry policy is just that if we do get that kind of upside growth that we were talking about at the beginning, then it increases the opportunity cost of industry policies. Yeah. Does that make sense?

So it increases the opportunity. So you're talking about industry policy elsewhere in the economy or industry policy for AI? Generally, including AI. So, I mean, there's a general issue in industry policy that as countries get richer and more productive, you increase the opportunity cost. and that's the point that I was making before. You know, if you're supporting businesses in this sector, you're moving capital, you're moving resources,

there's a higher cost to doing that because what you're doing elsewhere is already kind of productive. So to the extent that AI makes us really productive at doing a whole lot of other things, it may well make any sort of industry policy intervention a more costly thing to do. A couple of years ago, I had an assume Taleb on the podcast. And one of the most important life rules I've taken from

Teleb is that you can't reason properly about any domain until you know whether it's thin-tailed or fat-tailed. In a thin-tailed domain like human height, the average tells you almost everything. In a fat-tailed domain like wealth or pandemics or book sales, a tiny number of extreme cases drive almost all the variants. Charity, as it turns out, is fat-tailed. By giving to the best charities, you can often have 100 times the impact of the average donor. Givewell evaluates charities purely on cost effectiveness. Its top four charities, which include medicine to prevent malaria and cash incentives for routine childhood vaccinations, save a statistical life for somewhere between $3,500 and $5,500 US dollars. Effective Altruism Australia is how Australians give to those charities tax deductibly and with 100%

pass-through. Until June 30, 2026, Effective Altruism Australia is matching new donations up to $300,000. To donate, go to eaa.org.org.au slash Joe Walker. That's ea.org.org. slash J-O-E-W-A-L-K-E-R. So my sentence is that at the moment Australian policy makes a very exercise by this question of how do we capture some of the value of the AI economy and stop a disproportionate? amount of the profits flowing offshore. You know, they're thinking about preventing what Andrew Shulton calls the Uberization of the economy,

where in the last generation of kind of tech behemoths, like not many of the profits were captured within Australia. And if we simplify the AI value chain into like three layers of the stack, so there's like the hardware and infrastructure layer, then the model layer, and then the implementation and application layer. My sense is that the conversation in Australia is about how can we capture some of the value is focused on the kind of top and bottom of the stack, specifically data centers and applications. Does that check out with you? Yeah, I think that's exactly right. I think, you know, what's pretty clear, we're not going to be playing in chips. That is a very massive

economy of scale business and there's not many places that are doing that. Data centers, we are absolutely playing and, you know, we can talk about this. There are, there are good reasons for that. We're very unlikely to be in the kind of frontier model business. Again, it's the scale. But, you know, I think we could be very successful in the kind of AI application area, you know, just as we've had a lot of success in software as a service. True, true. So let's start with the application layer and then I'll come back to data centres. As you know, this is sort of a small hobby horse of mine at the moment. But on applications, one concern I have with the kind of, let's build it the application layer

vision is just, and maybe this is naive, but won't the major labs just leapfrog any niche applications that we build with increasingly powerful new foundational models? I'm just thinking back to like the early days of chat GPT where people. people would build all these wrappers and then with a new model release, they would just get sent straight to the startup graveyard overnight. Are there like principled reasons that I shouldn't be worried about this question? I mean, I think, like, it's what are the problems that those apps are trying to solve? You know, often it's kind of an understanding of way that businesses it's structured themselves and processes, you know, the kind of human bit that those apps go to,

which I still think the frontier models aren't always going to be the best for. So I suspect there will still be a role for them. It's a good question. But, yeah, I think, you know, it, you know, it's going to be if they can actually find ways to solve the very real world problem sitting there inside businesses or for individuals that the frontier models kind of can't jump to, then that's going to be where they create the value ad. Yeah.

I think it definitely will make picking winners more difficult. Absolutely. I mean, I think this would be an extremely high degree of difficulty thing to do in this space. Let's be honest. And then maybe the kind of totalizing version of the argument where it's just like, you know, almost everything in the application layer you build is like potentially vulnerable. Maybe that really relies on some sort of like transformative AI premise where it's aGI or something like that. So that's probably more speculative. Yeah, maybe it's an AGI.

That's right, where we're all, you know, building our own business applications and solutions using it because it's so powerful. but I do think that's probably a way off. Yeah. So one reason that we might be just naturally less worried about the Uberization of the economy with respect to LLMs is just that in contrast to the previous generation of tech behemoths, the LLMs aren't the network effects are largely absent from these businesses. I wouldn't say totally absent. And there are still some ways of potentially locking in customers like,

for example, if my LLM can sort of remember me in previous conversations and I can't kind of like export that memory very easily and transfer it to a different provider, then I might be locked in. But if customers aren't locked into LLN providers, that that might lead us to expect that profits are kind of competed down for those companies. Does that make sense to you? Yeah, I think that's exactly right. I mean, I think the network effects of things like social media platforms were overwhelming and in those kind of highly networked markets you get tipping.

And that's exactly what we saw. So you end up with kind of one or two big providers that have a lot of power. Here, I think, you know, you don't really have network effects. you do have some things that might push you towards lock-in and lack of ability to switch, but they're much less powerful forces than we saw of previous waves of technology. So I think that's exactly right. I think, you know, policymakers should be alive to the risks of companies doing things which could create stronger lock-in.

They should be alive to, you know, mergers and acquisitions. You know, that's something that I think with the benefit of high-ins, site. Some of the antitrust regulators in the US said, you know, maybe you wouldn't have let, you know, Facebook by Instagram, those sort of things with something's emerging. It can look low risk, but can create greater consolidation and market power. But I think, you know, based on what we know about the kind of economic dynamics of these models, there is less risk of sort of serious consolidation or less risk of kind of monopoly power than we saw in those previous generations of tech platforms.

And if that is true, then hopefully it would mean for Australia that more of the value will flow to the implementation and applications layer of the stack. That's right, that they wouldn't be extracting the kind of monoply profits or rents out of those players that come into the market at that layer. So another way to think about how much value will accrue to the implementation and application layer versus the model layer is just to ask what is the marginal value of intelligence itself? So, like, is intelligent, how much of a bottleneck is intelligence for firms? And if it's high, if the marginal value is high, then that would favour the model layer, I think.

Yeah, I think that's right. And obviously that will look pretty different across the across companies. Economy, you know, as we've been talking about with the labour market effects, you know, there are a whole lot of other things that aren't kind of straight intelligence related that go into value creation from different jobs. So that's right, to the extent that is the binding constraint that will favour those companies, but then the kind of competitive, the competitiveness of the market comes into play there.

So to the extent that a customer can still up and switch, you'd expect. some of that to get competed away. As a thought experiment, so take the Productivity Commission, because that's the organisation you know best, how big would the marginal product be for you at the Productivity Commission of moving from, say, today's models to, like, transformative AI models? Is intelligence the bottleneck for you? It's a great question, and it would... It would depend how fundamentally we wanted to embed it into our work.

So at the moment, you know, if I think about the way we're using AI, it's, you know, to help with our research, to help with coding and data analysis, to summarize market information, submissions, meetings that we have. So, you know, the kind of shift to transformative AI, you know, just makes those things a little bit better and a little bit faster. It's not the bottleneck. I guess the fundamental transformation would be if we had AI running the kind of end-to-end research process and reports and it kind of did all the things,

you know, asking the questions, going out, doing the research. somehow running consultation, coming back, forming recommendations, handing them to government. Basically, I don't exist. Not at my colleagues. So yes and no. I don't know whether that answers your question. But it goes to that kind of point we're making, I think. There's ways to use technologies which are about task-specific efficiencies and speeding things up,

and that's exactly how we're using it. And there are ways in which it's a total transformation of the way you deliver things. So the economist, Louis Garacano, who is the same economist who wrote the book, Messy Jobs, which you mentioned earlier. He has this idea that middle power should collaborate to build a CERN for AI to fund and build open source models that are pegged like one or two tiers behind the closed source frontier models. and the logic is that it puts like a ceiling on their pricing, it means that they can't tacitly collude because what prevents them from tacitly colluding because customers can always just opt for a less capable

but much cheaper open source alternative. And then that is going to let profits flow down to the downstream layer of applications and implementation. What do you make of this idea? Look, it's an interesting concept. I mean, and we've had this, you know, debate in other parts of the economy for a long time, you know, like whether governments should have their own bank or their own super fund, which, you know, creates a competitive dynamic such that others, you know, have a stronger competitor that they have to respond to. I think probably the kind of added argument here is, you know, what if it's, you know, what if it's, you know,

not just people kind of exploiting a strong competitive position, but they're acting in a way because of geopolitical reasons or other things that's kind of detrimental to countries' outcomes and national security. So this is a kind of safety net that we kind of have this, you know, neutral, global platform that people can access that isn't controlled by a company or a government in the same way. So look, I think it's an interesting proposition. It would be expensive. I think there's no question about that.

These frontier models seem to cost a lot to build, although China did there's, well, at least reportedly, significantly, more cheaply. It would require quite a degree of coordination and cooperation amongst middle power economies. So, look, I wouldn't discount it as an idea, but I think you'd have to do a lot of careful work and a lot of sort of thinking through costs and benefits. So on this question of diffusion, I've got like a double barrel question. Can you think of any historical examples where policy has made a big difference,

either positively or negatively, to the optimal diffusion of a new technology? And then does that have any lessons for AI diffusion? Yeah. So as I said, I think kind of diffusion is the main game here. and, you know, ultimately, whether we reap the benefits of AI is going to depend on businesses actually kind of adopting it and changing their practices to incorporate it. And so a good example in Australia in history is called Agricultural Extension Services. And so these services have been around and provided by government. hundred years. And this is like a very old concept, which is the idea you've got a whole lot of

farmers, they're very diffuse groups. So government would kind of go out to farms or meet with groups of farmers and, you know, talk to them about what they know about technology and ag, which could be, you know, new crop varieties, it could be pest control, it could be land management techniques, so that farmers have kind of the best information about how to employ new ways of doing things on farms. And those services over time have been judged to be, you know, really effective. It's actually shifting farmer behaviour. There's been some positive evaluations.

And, you know, one of the reasons people posit that Australia has got such a great record on agricultural productivity is that we've been really good at diffusing those ideas. And actually, when those extension services were really powerful was, you know, when they would sort of take information from farmers back to researchers. so they'd understand the actual problems on the ground on farms, research efforts would go into, you know, how do we control pest X? And then that information was then fed back to farmers again.

So you sort of had this kind of innovation loop. So, you know, I think what does that mean for AI? Well, you know, probably the biggest challenge with diffusion is going to be around business information and then management capability, the actual capacity to kind of roll out those changes. So you can imagine these sort of extension services which target small and medium businesses, the ones that tend to have the challenge here.

We see big business adoption is kind of powering ahead. They've got the resources and the capability to do it. But small and medium businesses may benefit from these types of services that government offers, which looks at their business model and, you know, works with them on the best way to implement that. There are lighter touch interventions, which is just kind of providing base level information across the economy, and we already have AI Center for Excellence, which is doing this. It says, you know, here's what an implementation plan looks like for AI technology. Anyone can go to the website and access that.

Extension services are a little bit more bespoke and hands-on. But, you know, that may be something that we want to think about when we're thinking about diffusion here. Yeah, it's super interesting. Do you have a sense for whether AI will diffuse, so even without some extra assistance from government, diffuse much faster than previous general purpose technologies? So, like, I was thinking about this yesterday. I mean, on the one hand, I have a sense that it could be the first self-diffusing general purpose technology. in a couple of ways.

One is that it's piggybacking on the internet. I think it's notable that ChatGBTGPT, I think, was the fastest growing consumer internet company in history, maybe 100 million users in its first two months, or something like that. But secondly, the AI can teach you how to use the AI. So in January, I used Claude to help teach me how to use Claude code, and now I use Claude code regularly.

But then, on the other hand, like that's maybe not what we mean by true diffusion. It's not, you know, in a really gritty way, kind of reorganizing organizational workflows and whatnot. So I'm not sure which way it kind of breaks, but do you have a sense whether it's going to be much faster than the internet, electricity? I think it will be faster, but not as fast as some of the technologists believe. So, you know, you've got this book on electricity in front of us, steam engine. Hold this up for the camera. Which looks fascinating, by the way. Networks of power.

But, you know, what we know from those early waves of general purpose technologies is, you know, it was decades sometimes, you know, more than 100 years before we saw the biggest impacts on total factor productivity because businesses had to kind of fundamentally re-engineer and reimagine and re-emvisage the way they did things to get the benefits. If we look at ICT, so computers, internet, it was faster than that, but still not overnight. And so one reason was the technology itself, as you said, it actually helps with the flow of information. The other reason is maybe we've got better as a society.

we've kind of learnt how to learn or learn how to do these things. And so I think those same benefits will apply with AI, but it's still only going to be as fast as, you know, organisations and individuals adopt it. And so, yes, we've got those kind of extraordinary growth statistics around individuals accessing it, but still, you know, only about 60% of Australian businesses say they're using it in some form.

And we know that, you know, for a lot of them, that's going to be fairly superficial uses. So, you know, I'm using it to write client emails or, you know, maybe I've got a kind of customer service chatbot up and running. You know, that's very different to the kind of transformation and embedding AI, you know, into the fundamental way that businesses work. And I still think, you know, it's not, it's going to be. years to decades rather than, you know, overnight. And that's why, you know, it's why you don't get the kind of, all the productivity benefits right away. It's why you don't get these kind of massive labour market adjustments.

It is just slower, I think, than people imagine. I think that's right. Another way to put it that's just occurred to me, AI might help with its own diffusion, but for that's happened, you need to be using AI in the first place. Indeed, exactly, exactly. And, you know, again, you know, we see it's much more common amongst exec leaders in bigger corporations. It's much more common amongst high education, high income workers and low education, low education workers. So, you know, I don't think it is yet as embedded in kind of every job and every business in the way that some people might assume it is.

Yeah. There's a question that's been bugging me for a kind of. couple of years now. How can Australia get a piece of the AI action? In particular, should we try to become, as Sam Altman has suggested we could, a data centre capital of the world? I recently worked through this question with Greg and EWen from E61, a non-partisan economic research institute focused on Australian public policy. It was a real joy sitting down with two of Australia's sharpest economists to apply some rigour to my nagging question. What we found surprised me.

In the long run, the financial returns to data centers may actually be modest, more like returns for electricity generation than for iron ore. But there is still a case for government support for data centers if they produce national benefits that aren't captured in the private returns. For example, we discuss how a compute industry could be important to Australia's sovereignty. We unpack these arguments and more in a new special edition for E61's plus 161. newsletter. To receive a copy of our essay on data centers and the compute economy, go to E61.in-slash-J-W-W-A-L-K-E-R. So some questions about data centers.

Your favorite topic. Maybe my favorite topic, we'll see. But do you have a sense for whether there are big economic rents on offer for countries who host data centres? I mean, I suspect not. You know, there's certainly conversations that Australia might be pretty well suited because we have a lot of land, we have a lot of potential for renewable energy. We've got a lot of wind and sun in particular. You know, we've got good institutions. we're a sort of stable democracy.

But, you know, I think broadly we will be competing in a, maybe not a global market, but a regional market for compute. It's not that other countries, you know, will not be able to build these types of data centres. So, you know, I think it probably will be a reasonably competitive global market. It's not, you know, iron ore or something where, We're one of, you know, few countries in the world that have high quality deposits and therefore we're able to exercise considerable market power. I just don't think probably plays out like that for data centers.

Yeah, I think that's probably right. So the argument for contemplating any possible government support for data centers is really like an externality's argument. I wonder, I mean, what, what things would have to be true, or what are some things that might have to be true, for you personally, to be satisfied or to support the concept of subsidising data centres? I mean, I think it would be a pretty high hurdle for me. So that compute was significantly constrained and without it, you know,

we couldn't get the benefit of AI throughout the broader economy. So, you know, there's some kind of bottleneck in compute that meant that, you know, Australian businesses were not able to use AI without a kind of, substantial government underwritten investment in data centres. And I just don't see that's going to be the world that we're in. I think there's pretty strong commercial incentive to build these things. And as I said, you know, there'll be a lot of other countries doing the same.

The other reason might be if there was a sort of security, kind of resilience argument that, you know, we don't want to rely on other countries for this. essential infrastructure. But again, you know, I just think to the extent, you know, we need to make sure that we have capacity for essential government tasks, you know, intelligence, those sort of things. Probably the commercial sector already has an incentive to build that here. And, you know, we see governments already kind of pay for a whole lot of things they need like I access

the secure network if I want to look at cabinet papers or those sort of things and that's it's not that government goes and builds that they just pay for it and I suspect the same happens here so if we weren't going to have any data centres but for

government coming in and supporting them and they needed to do that for a sort of national security resilience reason then that would make sense but I just think the market is going to deliver more than what we need in that sense. Yeah.

Let me add another rebuttal to the resilience rationale for subsidising data centres and tell me if this makes sense to you. So I was reading, yesterday I was reading the Productivity Commission's Guard Rails for Industry Policy Report, which was published last year. But there was one point in that that just got me thinking about this, which was, wouldn't we just be shifting the supply chain risk up the supply chain to like chips or something? If we hadn't built them yet, then potentially yes. If we hadn't built them yet. Well, so let's say, you know, if we identified this as a problem,

I mean, this is very hypothetical work because I said, I think the market's just going to sort this problem out. But let's say we... decided we're not going to have enough compute to do things that we think we absolutely must have as a country for our national security. And we then go out to market, to build our own data centres. But, you know, other countries are now kind of decided that they're going to hold up the supply of chips or whatever it is we need.

We're in that world. But if we identified as a problem, we build. the data centers now, we're fine. We've then got that capacity once it's sitting sitting on the ground. We've got the chips. We've got the chips. Yeah. Yeah. So it's only if both of those kind of emergency situations happened at the same time. Yeah, we've got. And we didn't already have the data centers here. The chips already baked in. Yeah. Okay. So there's like one other positive externality that I could think of, but it's a little more tenuous. And that's just that having a lot of compute, like we're talking in the order of magnitude of kind of tens of gigawatts,

not low ones of gigawatts. But that, particularly when applied to AI training rather than just inference, will give us the ability to, will give us leverage at other ways of the AI stack, especially for frontier models, sort of in a world where access to those models might be rationed. because they're, you know, ultimately will be importing those models from American companies and they're ultimately subject to the whims of US governments. But also we'll plausibly be able to get access to increasingly powerful models that aren't released to the public yet, like Claude Mythos preview and beyond.

I say this is tenuous because I think it really strongly depends on the particularities of any kind of contracts or agreements between data center operators, Australian governments and Frontier Labs. So I don't even know what this sort of looks like, but it just seems like it's possibly one other kind of positive externality in the possibility space. It's giving us some kind of negotiation power with either the companies themselves or the governments that are controlling the actions of those companies.

So I'm told that Anthropic is looking for its second country at the moment, outside of the US, whether that's Australia or Japan or somewhere else. I've no idea. But if there was some kind of MOU or bargain struck with Anthropic, which was okay, we're going to do, you know, I'm making these numbers up. But up to a third of your next however many training runs and in exchange, we would like access to the next versions of Claude Mythos preview to help with our cyber. security, when they become available, that's the kind of things I'm contemplating here. I, look, I can see what you're saying. I just wonder if it's, you know, if you're talking about that as a motivation for kind of government investment in data centers, I think it's a fairly long bow. A few too many links in the chain.

A few too many links in the chain, perhaps, yes. Okay, fair enough. Especially when I do think, you know, we will end up with a, a lot of capacity in any case because of the sort of market advantages that we have that we've spoken about. Yeah. So I was reading the government's expectations of data centers document, which came out, I think, in March.

And it's got five different expectations. Some of them are, like, pretty reasonable. It's like efficient water use and supporting the energy transition. There are others like providing, like, skilled jobs and apprenticeships. and also support like compute for Australian, like favorable terms for compute for Australian startups and nonprofits, which feel a little bit everything bagel to me. I mean, reading the document, it was reminding me a little bit of reading the U.S. government's notice of funding opportunity for new fabs under the Chips Act, where they were

just kind of layering on all these additional requirements. Are we being to everything bagel with data centers here? Look, I think there is a tendency towards everything bagel with almost everything these days. Look, I think, you know, as you say, the requirements around water and power, you know, I don't think anyone is going to disagree with. I think there's a genuine kind of community concern, particularly if these things are going up fast, if they start, you know, spiking the prices for electricity or water, you're going to see, I think, that kind of social licence degrade pretty quickly. So I think the government is right to focus on that and to say, you know, that we need these firms to kind of come up with their own solutions.

I do worry about putting a whole lot of other requirements on any sector beyond what is, you know, already there under the law around, you know, workers' rights and protections, around, you know, commercial agreements to supply to particular research organisations or centres. You know, the assistant minister in charge of, I think it's digital economy, it's called, Andrew Charlton's doing a speech today around data centres and the importance of community trust. And I understand that, but I think it's probably getting the kids. pieces, right, around water, around power, around kind of license within communities, rather than adding on a whole lot of other requirements, which is really pivotal for that trust piece.

Yeah, I'll have to read the speech. It does strike me if you by the very bullish arguments on the importance of data centres. Data center nimbusimmyism might become even more of a problem than residential housing nimbusimism in the next five to ten years. Gosh, that's frightening to think that that could be. the case, but I mean, I mean, there's a very real question around degree of kind of regulatory hurdles. And, you know, it is true that it's just hard to build things in this country, you know,

whether that is housing, whether that's renewable energy infrastructure, whether it's data centres, we do ask people to jump through a lot of hoops. And, you know, I think we should be looking to streamline that in a sense. sensible way, not necessarily remove protections, but to certainly to streamline processes as much as possible. And government's already done that to some extent through the changes it made to the Environmental Protection and Biodiversity Act last year, streamlining environmental approvals. But, you know, that's got to be an ongoing conversation for data centers, just as it is for all those other sorts of things we need to build more of.

Do you have any takes on this question of what if AI investment is a bubble? I mean, I think the, you know, this is very hotly contested at the moment, certainly in the US, and you've got people that feel very strongly on both sides. You know, certainly the kind of history of general purpose technologies is you do tend to get some bubbles and booms and busts before you find the equilibrium. You know, why would we be worried? well, if we were making all these kind of investments that were ultimately worthless, we might be concerned about that.

I think, you know, this is probably closer to, you know, steel boom and those things where, you know, at least we've got a productive asset at the end of it. So, yes, some investors might take a nasty haircut and, you know, that's bad for them. But as a country, if we're still got these data centres that we're going to be able to use at the end of it, it doesn't really have the same kind of negative productivity shocks that you might be worried about with a bubble that was producing something that didn't have any kind of inherent value or capacity at the end of it. So I want to finish on AI regulation and there's been questions about how we measure progress. But just before we do that, sort of cap off the AI and industry policy segment to sort of bring together everything we've spoken about, do you have a kind of gut sense for where in the stack the profits will ultimately flow? I mean, I think the chips are clearly. So I think the upstream, there's just only ever going to be a small number of firms. So I think they have a huge amount of market power.

So I suspect that's where a lot of the action is. You know, I hope we can have a kind of dynamic application layer where there's money to be made and Australian firms are a part of that action. But, you know, like anything in AI, I just think we've got to be pretty modest about forecasting where things go. And does it feel implausible to you that Australia could be one of the big countries in the value train. So that'd be like the US, China, Taiwan, maybe the Netherlands because of ASML and then Australia.

Look, I think, I mean, we have to think about it, I think, in proportion to our size. But, you know, as I said before, we have had, I think, kind of remarkable success in the kind of software as a service. area. Lascians, your canvars, your culture amps. And we've done really well. We have created these kind of ecosystems. We, you know, developed for the world from day one because the local market is small. So I think we have some of the ingredients for success.

But, you know, it's not going to be the same as, you know, having a chip manufacturer, I suspect. Yeah. So as a kind of segue into AI and regular. But the current most important binding constraint on training compute in Australia is copyright. And I was just hoping you could just, just for people's context, just like briefly explain your understanding of the mechanism by which our current copyright settings prevent training compute from being built here. So essentially they restrict the capacity of the trainers to pick up content and use that

to train off because it could be a breach of copyright law. Yeah. So that's the concern. And it goes kind of beyond, you know, what you might normally think of as kind of copyright type material. So, you know, something that someone's written in a book or that's in the newspaper, you know, potentially any use of, you know, the Productivity Commission website or, you know, scraping that material could all breach.

copyright laws. And potentially also, I mean, if the labs had to pay for that, then potentially also set like an international precedent as well, which might be a concern? That's right, although that is kind of the way the market, the market, I say, yeah, the market is evolving both here and internationally as we are seeing the labs now negotiating more with big copyright holders to pay. for access to their material and their content. Yeah.

So it feels like we're in limbo on the copyright stuff at the moment. I was saying Anthropics looking for its second country. Like, I don't know when it wants that additional compute to come online by, but maybe it's end of 2026, early 2027. It feels like really urgent that we have this very obvious barrier to training compute in Australia. And I just don't know whether we even like arrived at a solution yet. Yeah, we haven't.

But what I would say is it's challenging everywhere. So even though kind of on paper, our laws are stricter, it's a very hard issue for copyright laws to what extent AI training is a breach of those because copyright is all about kind of reproducing the material, like to what extent is training doing that. And so the US, I think there's some insane number of court cases. running that are live right now about this very issue. So I don't think it's fair to say that this kind of what you can do with copyright material is clear anywhere in the world at the moment.

So we grappled with this when we were, you know, thinking about productivity opportunities and data and digital last year and, you know, we flirted with different options. We put out for consultation the idea of, you know, should we have, have text and data mining exemption, should we have things? But where we ended up was actually, well, this is kind of being worked out globally right now. I don't think it's too much of an additional barrier in Australia and this kind of commercial relationships are starting to build up and form as well, which is picking up the part of the market that we might be worried about. So wait and see approach actually isn't crazy in that world where a lot of this is in flux globally.

So in the government's national AI plan, I noticed that they want workers and unions to have a strong voice in how AI is adopted across workplaces. Does that mean they're thinking about regulating AI as an industrial relations issue? Because it just seems like that could potentially really slow diffusion. Yeah, I would be concerned if they were thinking about it like that, but I don't think they are. So what they've said, which I think is right, and there's a sort of, I don't know what they call it a council or something that's been developed, which is bringing together, you know, business and government and unions to deal with these issues is, you know, really workers should be consulted on the rollout of technology. And I think, you know, that's largely required under IR law already. and also it's just kind of good management practice 101 that you should work with workers

if you're bringing in new technologies. You know, maybe there'll be some requirements for skilling and retraining under EBAs. So, you know, if augmentation is the big story, as we said before, we might want to give workers the chance to kind of build up the skills that they're going to need for those new types of jobs. And also, you know, some restrictions around,

digital surveillance and those types of, you know, uses of the technologies that could have a kind of negative impact on the quality of jobs for workers. So those are all, you know, I think pretty reasonable things that governments, you know, should be thinking about and businesses and unions should be working together on. What the government said so far, and I think is good, but is that's very different to kind of unions having a, or workers having a right of veto over technology, which says, you know, no, we can't use AI in this organisation.

And I would absolutely be worried about that because then potentially you are cutting off a lot of the potential productivity benefits of the technology. So at the moment, I think we're kind of walking the line in a pretty good way. But, you know, that's certainly a risk that we're thinking about that you might, at some point, people might be looking to move to veto rights. Scaling a business is hard enough already. So you don't want to be bottlenecked on things like compliance. And while achieving compliance can help you win bigger deals and to new markets and deepen trust with customers,

it can also cost a lot of time and money. By automating up to 90% of work involved in SOC2, ISO-201 and more, Vanta gets you compliant quickly. It also saves you money. A recent white paper by the International Data Corporation found that Vanta customers achieved $535,000 per year in benefits, and the platform pays for itself in just three months. A startup I worked at used Vantar and I found it really reliable. Over 12,000 global companies like Atlassian, Replit and Cursar also use Vantor to manage risk and prove security in real time.

My audience can get a special offer of $1,000 off Vantor at vanta.com slash joe. That's V-A-N-T-A-com slash J-O-E for $1,000 off. So in its report last year, the Productivity Commission argued that regulatory responses to AI should happen within existing legal frameworks and that AI-specific regulation should be a last resort. I'd be curious, just to play devil's advocate, what do you think is the most plausible reason why that approach could ultimately turn out to be the wrong one? I mean, I think it would be if AI was posing a lot of risks that sat outside existing legal frameworks. Right.

So, I mean, when we looked at this issue, what we saw was, well, when you think about the kind of potential harms from AI, you know, product safety, privacy, defamation, consumer harms. like a whole lot of things that we already have laws around. And it seemed to us a big risk to come in and say, okay, now we're going to have an AI act, and we're going to try and somehow directly regulate this technology, which presumably would be about trying to pick up those harms. But you kind of create a layer of duplication, complexity. We've seen it sort of done in Europe already.

the GDPR regime, which was around kind of privacy law, had really kind of slowed innovation, restricted access to products for consumers. They've actually interestingly just done it for AI more recently, and all the same concerns are there. So an approach where you kind of went to the existing laws and said, okay, what are the gaps here? Are they fit for purpose? There probably are new things and new risks that AI is creating that we haven't kind of got in the legislation at the moment.

let's fill those gaps seem to be the right answer. And, you know, it was already going on. TGA had already done a kind of gap analysis and proposed some amendments. Other regulators were doing it. We want to make sure it's done in a systematic way. The government said it will do that in a systematic way. But that seems to capture a lot of the action. But yes, if there are harms that sit outside those kind of standard regulatory frameworks,

then that's when that approach would have limitations. Yeah, so on that, is your gut sense that AI will turn out to be a quote-unquote normal technology? So it diffuses slowly and our institutions shape its course and maintain in control of the technology. Or how likely do you think it could turn out to be sort of an abnormal technology where it's less like a new tool and more like a new species, which is really fast developing and autonomous, because then obviously that goes to this question of whether there might be risks outside of the existing frameworks. Yeah, it's such an important question.

And so everything that I've just talked about, you know, approach, gap analysis, all of that is kind of AI as a normal technology. Yes, it poses risk, but we can kind of deal with those within kind of our standard regulatory frameworks. There is, you know, another very live, somewhat scary, conversation, frankly, that's going on about, you know, is AI godlike?

Does it pose an existential threat? Is it going to turn us all into paper clips that's going on? Which, you know, I think we should think about, we should pay attention to. Andrew Lee, a system minister in this government recently gave a very interesting speech about how economists should think about extinction risks more and weigh these into decision-making. And frankly, you know, when you have, you know, quite a number of credible people in this space saying, you know, we should pay attention to this.

I think we should. That said, is an Australian AI Act going to protect Australians from the existential risks of AI getting out of control? No. Like, this is going to happen over. overseas and we will get, you know, taken out in the crossfire, whatever it looks like. And so, you know, I don't see that, you know, putting in place cumbersome or complex or unnecessary regulation is actually not going to protect against that potential harm would be

the right way that we should think about it here. Essentially, it's a, well, it's a problem for regulations in the countries in which these models operate, but it's also a kind of global problem. We've just launched an AI Safety Institute in Australia that is working with AI Safety Institutes that exist in a range of other countries in Korea and Japan and Europe and Canada. And, you know, I think probably some kind of global coordination that is playing a role in proving the kind of transparent.

in auditing of these models that is kind of kicking the tyres on these risks of lack of alignment is probably the way we play the role. So Derek Thompson, co-author of Abundance, picked up on this kind of, is it normal, is it abnormal and said, why can't it be both? And I guess that's kind of how I think about it. I think to a large extent I think about it as a normal technology. I certainly think about the Australian regulatory response falling. into that category.

But yes, we should engage with the abnormal technology risks, but I think the right mechanism for us to do that is through a sort of playing a role in global cooperation rather than weighing down a domestic business with a legislative framework that, you know, frankly, is not going to have any impact on those kind of global existential risks. So my next question was going to be, if not an AI Act, what is the best,

way for Australia to get leverage over AI governance. So it sounds like the AI Safety Institute and that international cooperation is one way. Are there any others? I mean, it's sort of, like, I think there will have to be probably more formal regimes of international cooperation. Like, you know, climate change is probably the closest parallel to this. Like, the only way that we address things is through, you know, each country signing on and everyone's got an incentive to free ride.

So we probably need some kind of international mechanism. So I think, you know, playing a role as a good corporate citizen, developing expertise in certain areas is going to be the way that we have the biggest impact. Yeah. One of the things I've been meaning to look into, and I haven't looked into this in any detail yet is sort of like the non-proliferation frameworks. And then again, whether we could be thinking of uranium is some kind of analogy for compute. So I think our position as a sort of upstream supplier of uranium has enabled us to extract, like, governance concessions from countries we sell uranium to that go, even.

and beyond the kind of baseline obligations of the NPT. I want to look into that. I mean, again, it goes to that broader, more tenuous, positive externality around data centers, which we spoke about, but potentially there's another kind of route there. Yeah, I mean, it would depend whether, you know, we had a sufficient share of compute for it to be genuine leverage. And that's going to be the difference with uranium. again, you know, and you've got a small handful of countries that can supply that. Whereas I suspect we won't have anywhere near as much leverage from compute power.

So to finish, measuring progress, which is a really interesting topic. Actually, the question I want to ask first kind of segues from what we were just talking about, but that's if, so I read Andrew's speech, Andrew Lee's speech about the economics of extinction. I thought it was brilliant, like one of my favourite. speeches of his, just incredible clarity of thought. But the question that left me with is, if we are interested in the area under the curve, so like the stream of all future human welfare, and I am on the kind of the social discount rate should be like zero or pretty close to zero bandwagon. And so I think that that stream of future welfare is an important consideration. It is

morally important. And so if that is what we should be, if that is what policymakers should be thinking about, then our measure of progress should include not just the economic growth rate, but also some measure of the existential risk or hazard rate of new technologies. So I'm with Andrew so far. But then my holdup is calculating the hazard rate is an extremely non-trivial task. when you're dealing with nighty and uncertainty. So I don't even know where to, and indeed can often kind of backfire and lead to like suboptimal

policy because maybe you, you're overestimated and then you're too precautionary and you kind of foreclosed like a lot of potential growth. So I just don't even know where to begin with that. I don't know if you have any thoughts or if you're just in the same kind of nihilistic boat as me. Well, I mean, I agree with you. it's incredibly hard. And, you know, what people have tended to do was kind of go to experts and you get

this just, yeah, survey the experts. You get this like incredibly kind of broad range. Like, this has happened for AI, you know, I can't even remember the range, but somewhere between, you know, almost nothing and like 40% of wiping out humanity. I mean, it's pretty. That range tells you something. Yeah, exactly, exactly. You know, just high degree of uncertainty.

Yeah. I mean, people have played around your. kind of nuclear proliferation example before, there's the, is it the doomsday clock? You know, how many, how many kind of seconds to midnight? We are like sort of formulations that make it a bit more tractable, but it doesn't remove the underlying problem that you're talking about, which is just like, how do you actually inform that judgment? So, look, I think it's incredibly hard to do.

You know, kind of what I took out of Andrew's speech, though, is, well, let's say, you know, there's kind of any significant probability of P. Doom, we should be, you know, making sensible policy steps. Like, directionally, we should be investing more. Exactly. But, yeah, it doesn't answer the question of comp. It's kind of easy in this case, because in a way, Australia can't spend a lot dealing with. this issue, like the sort of interventions that we're talking about at safety institutes and international cooperation by their nature are almost certainly going to be worth it, relatively low cost. Much bigger question if you start thinking from the perspective of, say, the US and saying, should we actually just pause development of these models?

And that was, you know, that was a debate. Remember, like a year or two ago, Elon Musk was saying we should just stop and others were saying we should stop now, then you're dealing with these kind of very big cost-benefit trade-offs without a real sense of what lies on the other side of that ratio. Yeah. So in the AI as normal technology essay, which is like the big essay from early last year on which the Derek Thompson essay you mentioned was based, the two authors, Narayan and Kapoor, talk about treating the task of reducing uncertainty as itself, like a first-rate policy goal.

I wonder if that uncertainty about AI risks, that is, I wonder if you had any thoughts about that as a policy goal, and if it is a legitimate policy goal, is there any entity in Australia whose responsibility it is currently to reduce uncertainty about AI risks, or would we need some kind of new institution? I mean, I kind of think of the... It depends and the AI Safety Institute's a kind of new body. But, you know, if it is playing a role with these kind of questions of algorithmic transparency, so, you know, how are people actually using these tools that starts to give us a better picture of risk?

you know, is, you know, are people writing queries about bioweapons? What, you know, how are they kind of, and how does, and how are the AIs responding to that? What information are they giving them, those sort of things? So I think there is something around kind of auditing transparency of how these things are being used in the wild, as well as kind of how the technology is responding. That goes to this question of understanding how real those risks. are. So yeah, I think again, that's going to be part of a broader effort, which, you know, maybe kind of pushed by governments in U.S. and China or might come globally that we'll

have to go to some of those questions. So moving on from the hazard rate now and just focusing on the growth term in this equation, if the national system of accounts and our kind of standard measures of productivity fail to capture a lot of the progress from AI, as indeed they have to an extent for the progress from other digital technologies. Are there policymaking implications of that measurement problem that worry you? Do you have a sense of sort of like how big the measurement, the mism Measurement could be? And then does that potentially lead us to make, policymakers to make bad decisions or decisions that they wouldn't otherwise? have made. Yeah, so the measurement problem comes about because to the extent, you know,

AI products are free, like social media was free, we don't have good ways of capturing that in the national account. So you're creating this kind of consumer value that just isn't there because you've got a zero price. And this has been, you know, like a longstanding problem. I mean, the earliest example was actually kind of unpaid care work, which we've had Marilyn Waring, fantastic New Zealand economist, has been writing about since the 80s. So we've always known that GDP, 1, is not a measure of welfare because it misses a whole lot of things that are valuable now,

you know, including these new technologies if they're priced at zero. So yes, the measure. measurement era becomes more acute if this is a growing share of the economy. But what does it mean for policymakers? Like I think this has always been, it's not like people are kind of blindly optimizing policy around the GDP figure. It's just kind of not how policy works. You know, we kind of think about...

It's not like a speed limit in a car or something. Yeah, exactly, exactly. Like, you know, proper, you know, microeconomic reform, the bread and butter of this age. you're thinking about costs and benefits in the broad. You're not just kind of running at a GDP number. So it's a problem in the sense that, you know, the number that we kind of report and think about is probably less good as a measure of welfare than it used to be,

but I think we've always recognised the need to think more broadly than that. And, you know, even this government has sort of explicitly done that. I don't know if you've seen the measuring what matters framework that they put out. with the budget. So basically they said, well, you know, GDP is not the same as welfare. Why don't we kind of come up with a set of other indicators of things that people value and care about? And lots of countries around the world have done this.

It's a kind of dashboard concept. It's a dashboard concept. You basically get a broad agreement across countries within people. You know, what do we care about? Yes, we care about incomes and our capacity to buy things, kind of our GDP proxy. But, you know, we want good education. We want to feel safe. We want great healthcare system and health outcomes.

We want social capital connection. We want democracy. We want to have a say in who's leading the country, who's running country. So, you know, you end up with kind of a dashboard of different indicators that go to how well a country is performing. across a range of things that matter to people's well-being. What are some ways AI could change the economy such that labour productivity ceases to be a good enough measure of progress? I think, I mean, I think a lot of it will show up in total factor productivity. So you will end up with high gains there.

So it may be that kind of that's where the productivity action is going forward. We do tend to use labour productivity a lot, I think partly because it's more intuitive for people, partly because it comes out quarterly, whereas total factor productivity only comes out once a year. Why is that? It's a good question. That's interesting.

do? I don't know the answer to that. It's just, I've just accepted it as the way, the way things are. You have to get David Gruen on and ask you. Yeah, yeah, yeah. But yeah, sorry, I interrupted you. So, look, I still think those measures will be useful. I mean, the main issue is the ones that we've already talked about that we're kind of missing bits of the kind of output measure. Yeah. If you have a whole, you know, if there's a lot of kind of free products sitting in there, you might be kind of missing productivity gains. Yeah.

I guess if, um, if like the labour share goes down or something or if there's a lot of displacement, then because like, so, so the reason, my understanding is the reason labor productivity is a pretty good measure of, or a good enough measure of welfare at the moment is, um, incomes tend to move with... Yeah. With labour productivity. Yeah. Yeah.

But if the labour market is like not how incomes are being distributed in the economy anymore, then... Yeah, okay. Okay. So that's probably not the world I... You know, my best guess is that we end up in for the kind of long reasons we've already discussed. But let's say we're in the kind of tail scenario. and there's not much, you know, most people don't have a job, then your capital share is going to be much higher.

And then you've obviously got questions of kind of distribution. But yes, in that case, your labour productivity is not going to tell you much about how kind of individual people are faring. Yeah. So I think that would be the issue. No, that makes sense. So it seems that if, so if intelligence becomes abundant and really important and the constraint on intelligence is compute and the constraint on compute is energy, and then like energy is

kind of even more directly the binding constraint on progress than it is today. And maybe in that world, rather than talking about labor productivity or maybe. mainly about labour productivity, like energy efficiency is a really important concept, like units of output per unit of energy. Could you see that? Energy is your kind of production factor rather than... Yeah, yeah. So like the way today everyone reports on and talks about labour productivity in the sort

of national debate and the media, it feels like the conversation should shift to energy efficiency? I think it will be relevant. I mean, in a way, I hope it's not, because I, you know, really want us to be in a world of energy abundance sooner rather than later. We do have, you know, a kind of transition period, but I am kind of optimistic about the potential to increase energy supply over the long term. For the reason we talked about, we have abundant natural resources that we can convert into energy, but in a short-term sense where that was kind of your binding constraint. Look, I mean, it's certainly going to be relevant for thinking about policy.

It's going to be incredibly important for thinking about how you ration that scarce resource, whether that ends up being your kind of overarching macro statistic. I just don't know. Well, we better leave it there. There's been a lot of fun. It has been. We should catch up in 10 years and see... See how bad our predictions were.

What were we right about? What were we wrong about? It'd be interesting. Well, assuming we've both still got jobs. Yeah, you never know. Well, thanks so much, Danny. Yeah, thanks for having me, Joe. I hope you enjoyed this episode.

If you did, you can support the show by leaving a five-star rating on Apple Podcasts or Spotify or by subscribing on YouTube. Thanks to this episode sponsors, you can find them in the episode description. And thanks to Bill Manos and the Macon Fynos Foundation for their generous patronage of the show. If you'd like to become a sponsor or patron, you can go to jNWpod.com slash sponsor or email me at joe at jnwpod.com. That's joe at jnwpod.com.

Thanks for listening. Until next time, chow.