Telstra mass outage; AI and human rights with Edward Santow

In this episode, Sarah breaks down why we should all care about the mass Telstra outage even if we weren’t directly affected, why it’s political, and the case for a new legal obligation on telecommunications companies that requires them to deliver a reliable service.

Sarah then sits down for a conversation with Edward Santow, a former Australian Human Rights Commissioner and current Director of Policy & Governance at the Human Technology Institute where he leads a number of initiatives to promote human-centred artificial intelligence. Sarah and Ed cut through the tech spin with the facts about AI, how it intersects with human rights, and what governments need to do to regulate Australia’s data centre boom to ensure our environment, water and communities are protected. This episode will give you the information you want for the conversations everyone’s having about the AI boom right now.

‘Machines in Our Image: The Need for Human Rights in the Age of AI by Edward Santow and Daniel Nellor’. 

AI data centre inquiry submission guide.

Follow Sarah on:

Instagram 

Facebook

Youtube

TikTok

Substack

Transcript

SPEAKER_04 0:00

Hello and welcome to the Sarah Hanson-Young Podcast. I know a politician doing a podcast is a bit weird, but the world's changed and it's changing fast, and we've got to try and do things differently. We need to make politics a force for good, and how we do that is by having good conversations with good people, interesting people, and getting engaged. That's what this podcast is all about. We don't need to burn the place down, but we certainly need to renovate it. I just want to say thank you so much for all your feedback and engagement on the podcast so far. It's wonderful to get your messages and comments. People writing in and giving ideas on the issues and the topics you'd like to hear about. And that's what this podcast is all about. You know, as a politician doing a podcast, it should really be about engaging with you on the issues that you care about and you want to hear about and you want action on. So keep those comments coming, keep that feedback coming. Um thank you uh to Kathy for your uh advice about how to make this podcast go even better. And um and to those who um have complained about the F-bombs, um, sorry about that. You know, I'm I'm trying my best. Sometimes it just has to be said. And just before I get into the conversation that we're gonna have today with Ed Santo about the risks and benefits of artificial intelligence, we're gonna do my shy report, or not so shy report, a bit of a hot take and information on Telstra and the mass outage that occurred last week. And I know it doesn't sound sexy, telecommunications, but let me tell you, before you think this doesn't apply to you, maybe you're on Vodafone or Optus. It really does. Because this is how our country runs these days. Last week's outage caused mass havoc across the country. Millions of phones ran dead, hundreds of trains cancelled, morning coffee orders scrapped because people couldn't even pay with their mobile phone or even use their F-Post machines. And the most serious of all, hundreds of triple zero calls went unanswered. How can a system we rely on for so many things in our daily lives be so vulnerable to collapse? I reckon many of you listening right now would be shocked to know that the mobile phone industry operates basically unregulated. For example, there's no laws that your mobile phone company needs to meet to ensure that your phone actually has coverage and works. So you pay your bill each month to Telstra, to Optus, to Vodafone, and they don't even have to deliver you a reliable service. It can drop out a hundred times and they can just say, sorry, whoops, bad coverage. So I want this point to really sink in for you. These companies have no legal requirement to make sure our mobile phone works. And despite Triple Zero being the most essential, essential service, it's run by Telstra, a multi-million dollar company that's legal obligation is actually to deliver profits to its shareholders, not to deliver a public service. And that's a policy failure, plain and simple. This is why politics matters. When I hear people say, Oh, I'm not that interested in politics. Well, were you interested in the fact that you couldn't get to work last Wednesday because the trains were dead, or that you couldn't let the school know that you were running late for school pickup, or that you couldn't check in and let your mum know that you got home safe. It's a political choice that these companies run for profit and who market themselves as the backbone of modern life currently have no legal standards they have to adhere to. Now, the the rumors of why this Telstra outage happened and why the whole service came crashing down last week was because the company had put off upgrading a piece of equipment that would have cost them about 20 grand. Yeah. A piece of equipment that was due for replacement years ago. Now, the various investigations will bear this out. But if it is in fact true, then it really begs the question: how can a company that banked $2.3 billion in profit last year not keep its equipment up to date? And has ended up costing the entire country billions, potentially billions of dollars more because of the massive failure. That's why this week I'll be jumping back on a plane and flying back to Canberra, because I've called a Senate inquiry into this Telstra failure. I'll be chairing this inquiry, and for me, this hearing is about getting to the truth, holding Telstra to account, and making sure that we don't keep accepting a system that leaves Australians vulnerable. I think we should start considering whether we need to nationalise Triple Zero, built for the public interest, not the business model of the big telcos, who can pick and choose how much they want to invest in their systems. How is it possible that there's no legal requirement on these companies? This is political. It's a political choice to force these companies to put their consumers first and public safety first, rather than just their profits. It feels like everywhere you go right now, people are talking about AI, the power of it, how it's made, where it's used, what it's used for, the good, the bad, Australia's involvement. And it's one of the reasons why I've recently established a Senate inquiry into AI data centers. Because we don't know that much about what's really going on. There's these big companies who are pushing the Australian government to be able to build these huge factories of AI. And they want very little regulation. They're going to use a lot of power, a lot of water, they're going to take up a lot of space right across the country. And we know communities are asking, well, what's the deal here? What will this really mean? But of course, the bigger, even overarching question of all this is, and what will this AI be used for? Who will benefit? Which is why today's guest, Ed Santo, is such a perfect person to talk to. Somebody I've known for quite some time. He used to be the Human Rights Commissioner here in Australia. And he's now the Director for Policy at the Human Technology Institute. He is at the intersection of human rights, community, and the advancement of tech. Let's get into it.

SPEAKER_02 6:48

I was working as a human rights lawyer at an organization that's now called the Justice and Equity Center. And we had this service where young people could come with their legal problems to us and we'd give them free legal advice and representation. And this was in the early 2010s, and we just saw this growing number of young people, some as young as 12 or 13, who all had a remarkably similar story. So they were all kids from New South Wales who had had some encounter with the police. It wasn't necessarily anything serious. In fact, in most cases, it wasn't serious at all. They may have been riding on the train without a ticket, something like that.

SPEAKER_04 7:33

These kind of misdemeanors, like very small scale.

SPEAKER_02 7:36

Exactly. And then suddenly they found themselves on this list. And being on the list meant that uh the police would come and check on them. Um, but but often several times a week, um, sometimes between midnight and 6 a.m. in their homes.

SPEAKER_04 7:54

Come and knock on the door.

SPEAKER_02 7:56

Check that they weren't committing crimes. Um and so we we worked on this for for literally years, and we couldn't work out how these young people were being put on this list. Uh, and it actually took uh a series of questions from your now colleague, um now known as Senator Shoebridge, uh, in the New South Wales Parliament to actually get to the bottom of what was happening. And that was that um the New South Wales Police had this early algorithmic or AI system that was meant to identify young people who were at risk of kind of going into a life of crime. So this were 1,800 young people who were all on this list, and the system learnt wrongly that Aboriginal uh and um uh dark-skinned young people were more likely to commit crime. So it turned out that over 50%. It had a significant bias. It had a significant bias. So it was trained on the New South Wales criminal justice uh data system data set. And so that data set, you know, has like literally, you know, more than a century of bias and prejudice built into it. And so we had learned that, you know, Aboriginal people were more likely to commit crime. And so just to give you one stat from that, um, more than 50%, about 55% of those young people who are on this list were Aboriginal or Torres State Islander. Now, less than 3% of the New South Wales population is indigenous. So it had gone massively wrong. And so for me, as a human rights lawyer, I was outraged and I was worried. I was worried about how this new technology, which was explicitly introduced to make that decision-making process better, to remove human bias or human prejudice, it was having the opposite effect.

SPEAKER_04 9:53

Entrenching it.

SPEAKER_02 9:54

Exactly.

SPEAKER_04 9:55

And and without kind of accountability.

SPEAKER_02 9:57

Exactly. And so the police, you know, maybe to in their defense, they said, no, no, we built this system so that we would not have individual police officers working out who was, you know, uh uh at risk of uh of crime. We wanted to take that kind of subjectivity out of it. The problem is that AI is built by humans and it's trained on decisions that we've made.

SPEAKER_04 10:22

So it is On information that already exists rightly, wrongly. You know, if you think about the amount of history that AI has consumed already, written history, spoken history, um, recorded history. That's individual people's view of things. That's right. So of course, all of those biases are fed into the system.

SPEAKER_02 10:43

Yeah. So when you think about the history of the criminal justice system in Australia, going right back to the early days of the colony, we know that if you were Aboriginal, you are much more likely to be uh treated with prejudice. There's no other word for it, with racism. Um you're much more likely to have been convicted by an all-white jury who didn't have, you know, a sense of fairness, as we would call it, in 2026. And so um when you then build a system that learns from that really old data, yeah, then it's like reawakening a zombie. It it actually brings that those old perspectives that we have tried really hard as a country to move away from. It brings it back.

SPEAKER_04 11:31

So, I mean, this is such an interesting and kind of extraordinary example of why AI as it currently is is not perfect, is able to be fooled and to fool.

SPEAKER_03 11:45

Yeah.

SPEAKER_04 11:45

I I keep thinking about how it feels as though even the terminology of artificial intelligence is some kind of, you know, magic. You know, it's it's created up there in the atmosphere, you know, maybe it's, you know, the tech bros have kind of worked it all out. Uh it it doesn't exist except uh in that realm. When in fact it's not artificial, it's extracted. It's an extracted intelligence. It's taken from the history books, it's taken from the songs, it's taken from newspaper articles, it's taken from the things people have input. It's it's human intelligence that has been repackaged and supercharged.

SPEAKER_02 12:29

Yeah, that's right. Um, there was this really famous um piece in the New Yorker a few years ago, just after ChatGPT kind of went live for the first time. And the journalist referred to it as being like uh creating like a pastiche of um human creation. Uh and and that is what it is, right? Like it's not something that is kind of um bestowed on us by some more intelligent being. It's it's exactly as you say, it's it's it's our creation. We we've designed these systems, but we've also fed it information that we have created over time. And some of that is the best of us, some of that is the best of human creation, the the most kind of um soaring musical um outputs, those sorts of things. Um and some of it isn't. Some of it is the worst of humanity, and it's all in this messy pot together, and it's learning from both of those sources.

SPEAKER_04 13:27

Um we've had parliamentary um inquiries where submissions have used AI and been just plain wrong. Parliamentarians have gotten in trouble for sourcing documents that didn't exist.

SPEAKER_03 13:41

Yeah.

SPEAKER_04 13:43

So there's there's all these mistakes being riddled through the process. Um, but people are embracing this technology. It is designed to what what we're being told is to make life easier, make life more efficient. How do you feel about that part of it? Is that all just tech spin? Are the tech bros just trying to sell us a product that we don't necessarily need?

SPEAKER_02 14:07

Life would be simpler if it was all spin, to be honest. I I sometimes say that it feels like we're living in a fever dream. You know, there are these dystopian um realities that we're being confronted with, and they they really are dystopian. That that that example I was giving before of the um that it's called the suspect target management plan in the New South Wales Police. That's that is truly awful and dystopian.

SPEAKER_01 14:29

Yes.

SPEAKER_02 14:30

And at the same time, there are some wonderful things that are happening as well. Not not long after I had that experience with the New South Wales Police, I had this kind of opposite experience of of AI. Um, so much of my work in human rights has been with people with a disability. And um the something as as common as a smartphone, I'm holding mine right now, um, that can make a huge, almost life-changing difference if you are blind or have a vision impairment. And it can literally help you see the world around you. Um imagine that. Imagine, you know, um being reliant on other people to do something as basic as going to the supermarket, you know, that kind of thing. And now you can go to the supermarket and you can hold up your phone and it'll tell you what's around you. Like that is that that literally changes lives for the better. And so, you know, I I'm I'm asked, you know, sometimes, are you for AI or are you against AI? And it seems to me like such a weird question. It's like saying, Are you are you in favor of water? Like, yes, I'm in favor of water when I'm thirsty. We need it for you know drinking, right? But if I am, you know, in the middle of a a deluge and my house is like two meters underwater, you know, I'm not I'm not in favor of it. Like, that's just not how we think, right? What we need to do when we when we're thinking now about something that is so pervasive, so powerful as AI is we should be embracing what makes it good, the things that that can actually make our lives more inclusive, our community more inclusive and our lives better. Um and we should push really hard against those things that are terrible, that are make our lives bad. And I think the one, you know, pervasive myth that some of the tech bros are trying to tell us is that we have to accept all the terrible stuff in order to grasp the good.

SPEAKER_04 16:21

And that is that it's a trade-off.

SPEAKER_02 16:23

Yeah, that it's a trade-off. And that there's there's nothing more bullshit than that, right? Like it is completely untrue. There is nothing we we we don't say, you know, for something like um motor cars that you know, we should just accept that every tenth motor car is going to kind of blow up. Yes. We don't do that.

SPEAKER_04 16:41

Yes, or the brakes won't work.

SPEAKER_02 16:42

Exactly. We say no, you have to make sure that those systems are safe and that people can drive those those cars.

SPEAKER_04 16:50

There's seat belts, there's brakes. Oh, and guess what? There's road rules.

SPEAKER_02 16:54

Exactly.

SPEAKER_04 16:55

There's speed limits.

SPEAKER_02 16:56

Exactly.

SPEAKER_04 16:56

You put all of this regulation around it. And I guess this comes back to like what is the purpose of this technology? If it's to help humanity, then humanity needs to be centered in the technology itself. People need to be at the center of all of this.

SPEAKER_02 17:12

I had this, I had this experience a couple of years ago, and this sounds like a story that's sort of too perfect to be true, but it actually happens to be true. So um it was my birthday, and it also happened to be Melbourne Cup Day. And my then six-year-old son, Gabriel, was saying, What's what is the Melbourne Cup? Like, we're not a big horse racing family. Yeah. Um, and so he said, What is what's the Melbourne Cup? What is horse racing? I said, Oh, okay. Well, you have like a horse, um, and then you have the human, the jockey, and they're like a team and they're riding against other horses and humans. And he goes, Oh, yeah, yeah, dad, I get it, I get it, okay. And then we were talking a bit further, and it became clear that he thought that the horse was riding the human rather than the other way around. And we all had a big laugh about it. But then it became clear to me, oh my God, this is such a metaphor for the life that we're living at the moment, where we're asked, we're being asked as the humans to contort ourselves to the machines, to kind of change the way that we want to live so that the machines can kind of operate more efficiently. That is completely the wrong way around.

SPEAKER_04 18:20

Wow, and um I'm sorry, but you this technology is being built for us to use to help us. So, yeah, I mean the the machines are not in control here.

unknown 18:32

No.

SPEAKER_04 18:32

And if we're handing over control to the machines, I think we've got a problem.

SPEAKER_02 18:36

Exactly. And that but that that's the choice we have now. We we can continue to hand over that control and that kind of autonomy to machines if we want to, or if we don't make some really big decisions now. So we can we can still not choose to go down that path. But um, that's a choice that has like a deadline to it. Yes. Um if we kind of just fail to make decisions, then you know, the most powerful companies in human history will make those decisions for us and they won't be decisions that we like.

SPEAKER_04 19:10

Well, that's what a lot of the concerns that are being raised um with me, and I've just established this new parliamentary inquiry into the big data centers. Um precisely because I'm hearing people are like, well, hang on a minute, where is the community in this? At what point do we get to start regulating? What does this mean? How much energy? There is this kind of concern that if we don't do something now, it's going to be too late. And of course, data centers themselves seem kind of a strange concept. Like, you know, what are they? They're they're big buildings with a lot of computers that are computing, they take a lot of power. They get really hot, don't they? So they kind of they need water to cool them down. And they actually take up more land space than people realize. You know, when you save something to the cloud, it's not just stuck up there in the air, is it? It's like it's saved somewhere in some data center on the earth. Yeah. So it's kind of this we we're having to reconstruct the conversation. It's almost backwards in order to work out how we need to regulate it or how we want to have it controlled.

SPEAKER_02 20:25

Yeah, that's right. I hate the metaphor of the cloud because it suggests that our kind of modern data is just kind of hanging above us like rain or something. Um, just ready, you know, to kind of fall down when we need it. Exactly, yeah, yeah, yeah. Um, it's the exact opposite image for what the reality is. The reality of of how our you know, vast, vast stores of data are collected and then processed is in these enormous, almost unimaginably big data centers, hulking structures of concrete, very few windows, where you just have row upon row of servers that are kind of storing and processing information about you, Sarah, and about me and about all of us. And um, you're right, they they they they're so big, they're so energy hungry that the kind of unit of measurement for a data center uh isn't its physical space, but for how much energy it uses.

SPEAKER_04 21:30

Right. So when somebody says we've got uh there's a proposal to build a you know 200 megawatt data center, they're talking about energy, the energy it consumes.

SPEAKER_03 21:41

Yeah.

SPEAKER_04 21:42

They're not talking about how much data's in there.

SPEAKER_03 21:45

No.

SPEAKER_04 21:45

It's all about how much energy they're consuming. And that's a that's a shitload of energy.

SPEAKER_02 21:50

Yeah, that's right. I mean, when you think about what Australia has distributed across the entire country at the moment in data center capabilities. It's about 1.2 to 1.3 gigawatts, which is about sort of 15% of the state of New South Wales's electricity use. So it's not, it's, it's, it's quite significant, actually. It's significant already. Yeah, exactly, spread across the entire country. But there are plans in place to double that within the next year or two. And we're now hearing about plans for individual data centers that might be, you know, 800 megawatts or even a gigawatt big. And then there are some that are even bigger than that. There's one in Louisiana which is about four gigawatts in size.

SPEAKER_04 22:41

Four gigawatts. I I think, you know, I've spent a lot of time talking about uh energy and renewables versus fossil fuels. I mean, a a gigawatt is is about as much as you kind of your big coal-fired power station can produce. Yeah. Like so we're talking shitloads of coal-fired power station ver worth of energy.

SPEAKER_02 23:04

Yeah. The single biggest uh energy user in Australia is the aluminium smelter in Tomago in New South Wales, and that's not quite a gigawatt. Um and so if we're talking about single, you know, data centres that might be significantly larger than that, then it's a scale of energy use that Australia has never even contemplated in the past. And the timeline is is incredibly ambitious from the tech companies as well.

SPEAKER_04 23:35

These big, massive data centers that are being proposed around the country, that's not for Australia's use, right?

SPEAKER_02 23:43

So it's it's useful to think about particularly the AI part of the data center story in two categories. So if you are typing a query into one of the generative AI applications like ChatGPT or Gemini or Claude, um, we we actually don't have enough data center capacity right now to reliably process all of those. So a lot of those are already being processed in data centers in other parts of the world, particularly in the United States. And that's a problem because those data centers are under enormous pressure. And so the question that a lot of companies are asking is we want to be able to use these kind of um AI applications that we're spending lots of money on. Um, and if we can't have you know those those queries processed, we've got a huge problem. So uh there's already a pathway just to increase that that category, which is the kind of you know, consumer use, the company use, that sort of thing. But there's a whole other category of AI data center activity, and that is at the moment concentrated in two countries in the world, uh the United States and China. And that's how you build these incredibly powerful AI models. And that's on another scale entirely. Uh so to build one of these new models, which they do, you know, every few weeks or every few months, um, it is just this like unimaginably big task uh of kind of you know computer chips operating at at full pelt. And that's where we're talking about not just an extra one or two gigawatts, we're talking about five, ten, twenty gigawatts of additional data center capacity. And that is to kind of yeah, be one of those major players in the world that is involved in building these new AR positions.

SPEAKER_04 25:41

Yes, and they're the they're the facilities that we're now hearing from the government are, you know, the the companies are coming in and saying, hey, Australia, you've got space.

SPEAKER_03 25:50

Yeah.

SPEAKER_04 25:52

You've got energy. We don't really have that much water. I don't know why where they think all the water is going to come from, but anyway. Uh you know, let's do a deal.

SPEAKER_03 26:01

Yeah.

SPEAKER_04 26:02

Why do you think Australia is the place they want to come?

SPEAKER_02 26:05

Yeah, so they're they're thinking about those two things, energy and water, and they're thinking about two other things. One is the access to those incredibly powerful chips, because the United States restricts access to those chips to only certain countries. And Australia is one of the countries that is allowed to kind of purchase and operate uh those chips without any significant restrictions. Um that's one thing.

SPEAKER_04 26:31

That's kind of a trade arrangement or you know, a government-to-government deal.

SPEAKER_02 26:35

Exactly. And it's and it's it's for military and geopolitical reasons primarily that the United States imposes those restrictions. And then the fourth thing is just um risk. You know, we we're a long way from the world's kind of worst complex. Um the the the idea of um you know, which had previously been happening of data centers being established in the UAE, in Saudi Arabia, in other parts of the Middle East is much more difficult to come at now that there is active hot conflict in those countries.

SPEAKER_04 27:07

And indeed, and look what happens. I mean, Donald Trump, you know, wakes up one day and says he wants to bomb the place.

SPEAKER_02 27:13

Yeah.

SPEAKER_04 27:15

The rest of the world goes, oh shit, what do we do now?

SPEAKER_02 27:18

So Well, and and and in return, you know, Iran has actually targeted some of those data centers with its missiles too. So it is it is a much more dangerous part of the world.

SPEAKER_04 27:28

Volatile.

SPEAKER_02 27:29

Yeah. So I think what that kind of big number of potential countries in the world that could be housing these data centers, um, when you actually apply those four criteria energy, water, access to chips, and a relatively safe and stable country, it becomes a very small group of countries that could could actually do that. Um one of which has long been the United States, but they're now at the limit of what they can do. So there are lots of parts of the United States where the electricity grid has become really unstable because of the massive draw of energy from the existing data centers.

SPEAKER_04 28:04

Aaron Ross Powell And I mean that's fascinating, isn't it, that in the US itself that the that this is these centers are having I don't I don't even like calling them centres. They're like they're they're they're factories, they're kind of they're they're facilities, they're they they're doing things, not just sitting there.

SPEAKER_02 28:21

Yeah.

SPEAKER_04 28:21

It's actually powering the whole time.

SPEAKER_02 28:24

Um I like I like that word factory. I think that's a really good word for it because it is making stuff, it's there's real activity happening. Yeah. And there's a lot of kind of turbines whizzing and so on.

SPEAKER_04 28:35

And you hear people, I mean, so there's the concern that these big data factories are having on um the energy grid in the US, but also the community and communities where they're where they've had them housed or where they're being built or expanded are starting to go, well, hang on, what's going on here? And I mean the noise is just one thing that people are starting to raise. Um, you know, just thinking about the kind of the the whirling and the and the the the the the activity that's going on all the time. And I guess that's why now Australia is looking at this and going, well, hang on, what what are the regulations we need or why would we take this on? What's what's the what do we get out of this? I keep thinking of it as a um a new extractive industry.

SPEAKER_03 29:24

Yeah.

SPEAKER_04 29:24

It's just like mining, isn't it? Like we let the mining c companies come, dig up our coal and gas or iron ore, you know, like dig it all up, flog it off. We didn't tax them properly, we're not getting, you know, value out of the resource. It's Australian, but it's all being exported overseas, and then we have to import it back in a better product.

SPEAKER_03 29:47

Yeah.

SPEAKER_04 29:47

What do we what have we got out of this? Like we've made this mistake already.

SPEAKER_03 29:52

Yeah. Yeah.

SPEAKER_04 29:53

Why would we let another extractive industry come in and take us for a ride?

SPEAKER_02 29:58

I think that's right. And I think we need to be uh more confident as a country about um the leverage that we have and also think really uh carefully about some of the best innovation decisions we've made in history. What we've actually said is um we're not just trying to build something for the narrow benefit of a of a very small number of people, we're trying to build something over the longer term. That's what Australia is really good at. Um and so when you when you look at some of the the headline figures, it's incredibly appealing. So the kind of rough um rule of thumb um when you're building a data center is that for every gigawatt of data center that you're building, it's about $30 billion, billion with a B. That's a lot of money.

SPEAKER_04 30:47

That's a lot of money.

SPEAKER_02 30:48

But here's here's the thing, right? Um, 80% of that cost is actually the chips. These are these these incredibly powerful that we don't own. That we don't own. They get manufactured here. So 80 cents on every dollar immediately leaves Australia, right? It just sort of, you know, out out the out the door straight away. And so, okay, so of that $30 billion, 80% is actually a bit of a mirage.

SPEAKER_00 31:16

So you're left with 20%, but 20% of uh $30 billion is a lot of money, right?

SPEAKER_02 31:22

And so I think what we need to then say is really clearly, well, how are we going to um d drive something and build something that is genuinely of value to the community as a whole? And um, how are we actually going to front up in an adult way to those challenges, particularly of energy and water and and a s that sort of sense of community safety around um where those data centers are rather than just kind of s thinking, oh well, she'll be right, mate. Yeah, yeah, exactly.

SPEAKER_04 31:54

Don't worry about it. Or the tech bros, oh yeah, yeah, they'll sort it out. Yeah, they've got it all underhand. Yeah, you know, it's fine.

SPEAKER_00 32:00

I think and I think They'll look after us.

SPEAKER_04 32:02

I mean, if your solution to your problem is that Elon Musk will look after you, I think you've asked the wrong question and you're barking up the wrong tree.

SPEAKER_02 32:10

Yeah, I think that's right. But but it it's striking to me that we haven't had that kind of public conversation about what what does this actually look like, right? Like we're yes, we've all we've all seen those headline figures of big investment. Um and I'm not saying that that's unimportant, um, but these are not big job creators. After you've built the the data center, um you don't have huge numbers of employees operating. It's not like the factories that we used to have, you know, in the kind of earlier parts of the 20th century that built cars and those sorts of things. This is a completely different scenario. And so what we need to do is have an open conversation in the Australian community about what does driving real value look like and what does building in a way that actually increases our renewable energy industry, um, that preserves water, that protects communities. Like, you know, we need to put the tech companies to proof to actually show how that can all be achieved. And if it can't, I think we also need to be adult enough to be able to say, well, nothing. Not here. Yeah.

SPEAKER_04 33:17

Not here. Yeah. I that that takes some strength and some spine from politicians, though, because it's so easy, I think, to to have an industry come in and promise the world $30 billion investment. Well, you've just explained to us it's not $30 billion. Uh oh, it's gonna create all these jobs maybe for the construction of the facility, the factory, the data factory, but they're not factory workers that are in there, they're computers.

unknown 33:48

Yeah.

SPEAKER_04 33:49

So it's not long-term workforce kind of investment. Energy, who the hell knows exactly how that's gonna happen? I mean, the there's a data center that's planned for South Australia at the moment. And correct me if I get the terminology wrong, but um uh eight hundred megawatts, which is more that's so that's how it's measured.

SPEAKER_03 34:12

Yeah.

SPEAKER_04 34:12

800 megawatts, that's more power than the whole of Adelaide.

SPEAKER_00 34:16

Yeah. Yeah.

unknown 34:17

Right.

SPEAKER_00 34:18

It's it's it's almost impossible to get your head around, isn't it?

SPEAKER_04 34:20

And and and you know, like we we're in the middle of trying to build out our energy grid so it is sustainable, so it's relying on renewables, so it's reliable for the community. And then, you know, the tech companies are just kind of gonna come in and plug into our grid.

SPEAKER_02 34:40

Yeah. Well, I mean, I think what they've been saying so far is that they want to build new uh energy generation that won't be a kind of a net draw from the grid, and they really want to focus on renewables. So what I would say is let's take them at their word, because um that's that's a great commitment, um, but to actually make good on that is difficult. So when you look at some of the new built um data centers in the United States, they rely really heavily on gas power. And if you've ever, I mean, Australia has the good fortune of not having a lot of um gas power generation, but if you've ever been near one of those, it is incredibly noisy.

SPEAKER_03 35:22

Yeah.

SPEAKER_02 35:22

Um, that's the thing that's making all the noise. Um, but also that is not renewable, you know. Natural gas is not renewable. No. Um, and so it's it's one thing to say, yeah, we'd we would prefer to build more renewable. Well, you know, I would prefer to be, you know, able to run the hundred meters in ten seconds. Yeah. But that's not gonna just be self-executing. Yeah, yeah, yeah. So you've got to find um, I think, uh a way to actually hold companies to their commitments.

SPEAKER_04 35:52

And not a, okay, we'll build it first and then we'll come we'll come with the the renewables after or the water solution after.

SPEAKER_03 35:59

Yeah.

SPEAKER_04 35:59

Because it sounds as though a lot of these companies are promising to get their builds done very quickly. And you said an extra gigawatt of data capacity in the next 12 months or or a couple of years. And that's pretty fast time frame.

SPEAKER_03 36:17

Yeah. Yeah.

SPEAKER_04 36:19

Wh th that renewable energy is not going to be online by then. Well doesn't seem really realistic, does it?

SPEAKER_02 36:25

We we've we haven't been able to achieve that sort of speed for building massive projects in Australia in the past. So maybe this will be the one where we do. But you know, you've got to you've got to have a realistic plan.

SPEAKER_04 36:38

Yeah. Feels to me, you know, you're saying there's a we're we're at a point right now where we have to we we can't just close our eyes and and and and hope somebody else sorts it out.

SPEAKER_03 36:49

Yeah.

SPEAKER_04 36:49

Uh it there is this is on, the pressure is on, the tech companies want to come here, they're pressuring our governments, they're promising big things. I mean, part of me thinks if it sounds too good to be true, it probably is. History usually shows that, but whatever. Uh, we've really got to kick the tires on this before we sign everything over. And that's part of the inquiry, the Senate inquiry that we're running. But uh I I guess my what I'm hearing from you is this is the moment to take stock and to say we can actually make a choice. We it's not we don't just have to blindly, you know, sleepwalk into this decision. We actually can make some decisions. What regulations do we need? What are the commitments that need to be in place? What is the community going to get back from this? And right back to your first point about what's the purpose of it? Is it is it about helping humanity, or is this going to actually divide us more or to create more problems? And who's accountable for that if that's what if that's what happens?

SPEAKER_02 37:58

Yeah, I think that's right. And I think what it, you know, what a mature country does is it doesn't just say, I've got to accept this sort of gift horse, you know, um, or I've got to reject it. They'll say, I'm willing to accept it, but only under these conditions. Um and that's what you know, having a kind of a mature debate will actually bring to the fore. I know what for me, um, you know, I'm I'm a lawyer, but before I was a lawyer, I was a human too. You know, I I know what for me kind of makes sense for me and my family, but you know, I I shouldn't have any greater say than anybody else. But for me though, the the most important things are, you know, I've got young kids. I don't want um my kids to grow up in a country that is, you know, sort of materially worse, like, you know, than the country I grew up in. I don't I don't want it to be dirtier, more polluting, you know, um with less access to water, all of those things than than than I experienced.

SPEAKER_04 39:00

But you'll have fast internet.

SPEAKER_02 39:01

But with fast internet, exactly, yeah. But I also don't want to I also don't want a community where we're not thinking deeply about the things that we're making, right? Like you, I mean you you mentioned earlier, you know, in in previous mining boons, we we'd kind of extract the iron out of the ground, we'd sell it to another country, and then they'd sell us back a car. A car, that's right.

SPEAKER_00 39:24

So we what do we do that for?

SPEAKER_04 39:25

You know it's that traditional kind of idea of well, how we how do we value add to the resources we've got rather than just flog them off?

SPEAKER_02 39:33

Yeah, that's right. And my my colleague, um Professor Nick Davis, says, you know, that there's a risk that the decisions we make here on data centers would be like the next kind of strip mining boom. But he's also really careful to say there are opportunities with with data centres, but those opportunities really rely on us putting in place very strong guardrails to make sure that we um were able to address those risks.

SPEAKER_04 40:02

Some people are calling for a um a moratorium or a pause on the opening of these massive centres. And you know, they're called hyperscale, which just says everything to me. That we should put a pause on these big ones because we don't have all this stuff sorted out. That doesn't mean we don't look at the advantages, but to your point, we might have to say no to some of them.

SPEAKER_02 40:26

I think I think if you're if you're in a negotiation from as strong a position as I think we are in, then you don't go into saying, look, the answer is yes, but please will you give us some other things?

SPEAKER_04 40:38

That's right.

SPEAKER_02 40:38

That's a crazy thing to do, right? Like it's it's much more sensible to say, look, you know, all the options are on the table. One of those options is yes, another option is no. Um, and it's only going to be a yes if we can satisfy the things that are most important. Um and uh I think at the moment we're we're not doing that.

SPEAKER_04 40:59

Back to the kind of AI itself, there is a lot, increasing concern of people are worried about how AI could be used. And you know, your example of the the list of of young people and and kids. It's not something that could happen. It did happen.

SPEAKER_03 41:16

Yeah.

SPEAKER_04 41:16

It did happen.

SPEAKER_03 41:17

That's right.

SPEAKER_04 41:18

Um who where should people go? There's so much information about AI, what this means, what it doesn't. I mean, if you ask AI about AI, that it comes up with some funny answers. Um, prefers not to talk about itself.

SPEAKER_03 41:38

Yeah.

SPEAKER_04 41:40

We need trusted information for people because we are in this crazy world where technology is actually supercharged misinformation and disinformation, and some of that deliberately. Yeah. Some of that to serve the purposes of the vested interests.

SPEAKER_03 41:56

Yeah.

SPEAKER_04 41:56

So I'm just thinking about people who are listening to our conversation today, like, where should they go to get good trusted information?

SPEAKER_02 42:04

I think um if someone's trying to sell you something, then they're in a reasonably good position to be able to talk about all of the great things about it. And so there's nothing wrong with listening to, you know, a Sam Altman or even an Elon Musk when they're trying to tell you all of the best things about the products and services that they're making. But you wouldn't go to those people as your only source of information. And there's there's actually been some fantastic um filmmaking, but also some really good books uh that that actually that sort of give you a much more kind of independent view of how AI operates. Um, one of the books that was most influential on me was by an Australian, um, an amazing woman, Kate Crawford, who's uh been based in the United States for a long time now. Um her book is called Atlas of AI. And one of the things that it does, um, I think particularly well is that it makes the point that you're making, Sarah, which is that AI isn't some kind of you know, bit of fairy dust that kind of hangs above us, like the kind of mythical cloud. It is very, very much grounded in reality. So she she starts by talking about um how um you know the the mining of rare metals, um, the kind of extraction of of coal to create you know dirty electricity, all of those sorts of things, um, that they're really, really central to how AI is built. And then she talks a bit about how it operates. And she's not, you know, anti AI at all. She's she's someone who has basically given her professional career to really understanding how the system works. And I think unless you understand how the system works, you're really going to struggle to understand um how we can make good like decisions as a community. About what we should embrace and what we should push back on. It's a bit of a kind of a cheeky thing to do. But a couple of years ago, I wrote a book myself with a wonderful colleague called Daniel Nellor. And what we tried to do was look at artificial intelligence from the perspective of people who care about human rights. So what are the things that really should matter to us if what we want to do is kind of, you know, protect people's dignity? That that I think is another way in. I know that, you know, human rights is kind of the lens through which I tend to look at problems. It's not the only lens. But I think it's a useful one. Because I think what it what it reminds us of.

SPEAKER_04 44:42

If human rights aren't the basis of how we engage with this technology, then why the fuck are we doing it?

SPEAKER_03 44:49

Yeah, yeah.

SPEAKER_04 44:50

What what what are those things, Ed, that we should be saying need to be on the table? What are we looking for? What are those markers? What should governments not sell us short on?

SPEAKER_02 45:01

So there are three key things that I think governments need to be laser focused on. The first is thinking about like kind of the physical manifestation of AI, which is mostly what we've been talking about today. So making good decisions about data centers, the energy and water use, all of that sort of stuff. So we've covered that. The second is we know that there are some ways in which uh AI can be used that actually cause harm. And most of our laws are kind of technology neutral, which means they apply to all technologies and none. But the problem is sometimes we don't enforce them equally. So I'll give you, I'll give you the example against myself. You know, when I was human rights commissioner, if we'd, you know, got wins that there was a crusty old bank manager who um thinks that women make bad customers and doesn't give any loans to women, we would have been all over that bank manager by the end of the day, right? Like we would have absolutely been laser focused on addressing that. But we know that's actually not the main problem today. Um we know that you know banks use AI to make home loan decisions, but actually they have that problem because those AI systems have been trained on historical data sets, and decades ago, women found it much, much harder to get a home loan. So those systems, wrongly, are kind of still making it hard for women to get a loan. Instead of saying, well, that's clearly unlawful, that's clearly discriminatory, we've got to take that person to court. Too often we kind of scratch our chins and we kind of go, hmm, what an interesting ethical problem. It's not, you know, I think regulators and the regulatory ecosystem that supports regulators have to be crystal clear. Those things, when they're unfair, when they're a breach of someone's you know, legally protected human rights, we need to absolutely enforce the law. Okay, so that's the second thing. And then the third thing is there are some gaps in our law. You know, privacy law is a really good example, right? Our privacy law was drafted mostly before the rise of the internet, let alone AI. So most countries in the world have actually modernized um their privacy law uh to make sure that people are protected. We haven't. We've been really slow to do that, even with commitments from you know, more than one attorney general, more than one Australian government.

SPEAKER_04 47:25

It's just one of those issues. I mean, I've been in parliament a long time now, and it's just one of those issues that constantly gets kicked down the road. Oh, maybe, maybe this time, oh no, no, it's too hard. No, maybe no.

SPEAKER_02 47:38

But you know what? That's having real-world consequences now. So just over a year ago, there was this really interesting investigation about how Facebook was training its generative AI models on their Facebook users' data.

SPEAKER_04 47:55

But actually Unbeknown to Facebook users. A lot of that.

SPEAKER_02 47:58

But most people had no idea, right? But the interesting thing was that they weren't doing it on all Facebook users. So for Facebook users in Europe, for example, where they did modernizate, they had modernized their law, they said no, we're not going to trade it on them. Even in countries, you know, that are a long way from Europe, like Brazil, they were not using Facebook users' um personal information to train their system. But you know, a country where they were using Facebook users here in Australia. And that's because we just hadn't modernized our laws to actually kind of basically set a boundary and say, no, like companies shouldn't be able to just come in and steal people's personal information and and and use it for their own purposes.

SPEAKER_04 48:40

You're at the epicenter of this. The way technology is being used, misused, the impact on humans, it's so big. Like, what gives you hope about trying to get this right?

SPEAKER_02 48:53

Um, so I'm a I'm a human rights lawyer, which means by definition, I'm an optimist. Um, and I do think there are some things to be optimistic about. The thing that really gives me hope is um there are some companies here in Australia who've kind of played to our strengths. Um, so Australia's been at the heart of geospatial mapping using AI, where you get like this extraordinary picture of, for example, the aftermath of a bushfire. Uh, and you can kind of get help to those animals that are literally dying in ways that we couldn't do that before. These are really good uses of AI. And what appeals to me about it is that it takes a genuine problem, one that we can all agree on is something that we should address. It uses technology in a very targeted way, and so I think it's those sorts of projects that are really thoughtful, that are really considered. Those are the sorts of things that give me hope.

SPEAKER_04 49:54

Well, from one optimist to another, um, here's hoping. Thank you so much, Ed, for coming in today. It's been fabulous. It's been an absolute pleasure. What I thought was the most poignant of what Ed said was that we had to be awake and aware to this pervasive myth that we should accept the terrible stuff in order to grasp the good, you know, the trade-off. And I mean, what a gaslighter argument that is. What an absolute play. Pretending that somehow, yes, this bad stuff will happen, but you there's all this good stuff, and you've just got to take the good with the bad. We don't actually have to do that. And in fact, as politicians, as government, as decision makers, our job is to expose that and to say, no, no. We will decide, the people will decide what the trade-offs will be and what we're what we're prepared to accept. And I thought that story around, you know, the the horse becoming the jockey, uh, you know, the machines running the place as if we're you know servants to them. No, it's got to be delivering for us, and we've got to know what the real costs are. Thanks for tuning in, and I would genuinely love to hear your feedback. I want to hear what you're liking, what you're not liking, what you'd like to hear more or less of. And if you hit the follow button on the podcast and leave a review wherever you're listening, uh, I'd be very grateful. Thanks, team.