Ep.75 From cautious to curious: the future of AI in water

Ryan Signor Ryan Signor
Industry Director, Water – Australia
Dave Mackenzie Dave Mackenzie
Group Director, Digital & AI
Amanda Lewry Amanda Lewry
General Manager for Growth, SA Water
Karlene Maywald Karlene Maywald
South Australian Ambassador for Water & Chair, WaterAid Australia
Pierre Pauliac Pierre Pauliac
CEO of Global Water, SUEZ
26 June 2025
19 min

Maria Rampa: Hi I’m Maria Rampa and welcome to this episode of Engineering Reimagined.

Imagine a system that can predict a burst water pipe before it happens. One that automatically isolates the affected area and reroutes it to quickly minimise disruption and downtime. With artificial intelligence this scenario is likely just around the corner.

The potential for AI in the water sector is enormous. Yet, compared to other industries, water is moving cautiously. While AI already balances energy loads in smart grids and enables autonomous drilling in oil and gas, many water utilities remain hesitant.

And that caution isn’t without reason. Water is local, for the community, it’s complex, and it leaves very little room for error. But Europe is beginning to show what’s possible – and what’s scalable.

In today’s episode of Engineering Reimagined recorded live at the Ozwater’25 conference, our guests explore what it will take to unlock AI’s potential in the water industry, and why now is the time to move from cautious to curious.

Moderated by Aurecon’s Water Market Industry Director, Australia, Dr Ryan Signor, today’s guests include South Australian Ambassador for Water and Chair of WaterAid Australia, the Honourable Karlene Maywald, the CEO of Global Water at SUEZ, Pierre Pauliac, SA Water’s General Manager for Growth, Amanda Lewry and Aurecon's Group Director for Digital and AI, Dave Mackenzie.

I hope you enjoy this fascinating panel discussion.

+++++

Ryan Signor: Welcome to this discussion on one of the most urgent and compelling intersections of our time, which is the intersection of artificial intelligence and its responsible use in critical services sectors like water. I want to share six key principles that will help frame the conversation for today.

The first one is the core tenet for today. In a sector as critical and important as the water utility sector, where the obligations are on being able to serve as stewards and harness all of the benefits and treat a critical resource and a series of assets that enable it to be provided to where it needs to be, to uphold the values that it supports, is a sector that is too important to fail. And it's the obligation on the sector to look after this public good in the most responsible way that it can. And so, the responsible use theory is one where we say that AI in the water sector is not only a matter of constraining some type of uses or applications, but also one of harnessing its great potential.

The core principle is when we need to be harnessing power to give this commodity and good to the people who need it and the places that need it, the best we can, is the unethically thing not to fully embrace AI in the most powerful and impactful way that we can.

The second principle is more of a reminder. It's the fact that AI is not something in the future, it's already here. AI is behind predictive maintenance in mining. It's around delivering autonomous drilling in oil and gas and energy load balancing in smart grids.

The third principle is a provocative one. Research will show that both in Australia and globally, the water sector is a cautious one, conservative and in some ways quite constrained in its AI adoption. Comparisons to other sectors have been made and the water sector continues to be seen as a laggard, both in the pace with which it is up taking AI for its uses, but also in the ways in which it is imagining what AI can do for it. Some of that caution is justified because water is deeply local, it's complex and there's limited room for error, but the opportunity cost of this hesitation is real and it could be growing. So, I encourage us all to ask, what might be holding us back?

The fourth principle is around knowing your AI. To be able to have a discussion on the opportunities and risks that it may present, we need to understand a couple of basic concepts, broadly we can talk about cognitive labour AI and AI agents, and this leads us to our next point. The evolution of agents is fast and rapid. Most water organisations are only scratching the surface of what AI can do, imagine a system that could predict a burst pipe before it happens, automatically isolates the affected area and reroutes flow, all without a human in the loop. Is that something we want to aspire to? And do we have the strategic courage to move beyond insight into action?

And finally, principle number six is about understanding the types of AI risks that can manifest. We can't talk about action without talking about risk. So, let's talk about micro risks. These are internal to systems. Data quality, algorithmic bias, system failures. For example, if an AI model misidentifies a water quality issue and sends false reports that could cause real harm without a human nor a corrective action in the loop.

How do we ensure that AI systems are pursuing the right goals? When humans are out of the loop, how do we verify that agents are acting in the public interest and not just optimising for efficiency? Second, malicious risks. These are external risks. AI can be hijacked. Deepfakes, sabotage or adversarial attacks on critical infrastructure can happen.

In 2021, a US water treatment plant was targeted by hackers trying to poison the water supply by adjusting chemical levels. We have to be vigilant to these types of malicious risks and security by design is not optional. It's going to be imperative.

And finally, macro risks, which are systemic. AI may lead to over-reliance on automation, loss of human expertise or shifts in power structures. If critical decisions about water or other things are made entirely by machines, how do we preserve transparency, accountability and public trust? These are not reasons to delay, but they are reasons to act wisely. So they’re the six principles that's going to create the context for the panel discussion that’s going to follow. How can we unlock the power and perils of AI? How can we ensure that it's serving the public good and creating the future that we hope for? Karlene, is there an obligation on this critical sector to lead into an AI future?

Karlene Maywald: I'm going to answer that by taking us back in history. When I did year 12, we were not allowed to use calculators in the year 12 exam because the world would end if students used calculators and didn't use their brain. Things have changed a little bit since then, and the world hasn't stopped. Secondly, when the internet came along, the world was going to end. This internet was going take over our systems. It was going destroy us. We embraced the internet. What I'm trying to highlight here is that we have gone through so many different changes in our industry over a very long period of time and we have been able to embrace those changes by being diligent and focusing on what we need to do. AI in my mind is no different to that. And we are not the only ones that are grappling with it.

When we have made significant changes in the water sector in the past, we've not set back and let the same people who've always done things the same way do the same thing. We've actually embraced the change by enhancing our workforce with the skills necessary to approach it. When we went to a customer centric and driven approach to our water utilities, we changed our staffing mix. We are going to have to do the same with AI, and we're going to have to bring on the skills that are going enable our workforce to embrace the issues and the challenges that we're going to be faced with AI. Because whether you like it or not, it's on their phones already. They've got them on their home systems. And so AI is with us, and it's how we manage it, and how we embrace it, and how we bring the skills on board to assist with that.

Ryan Signor: Amanda, what's the utilities' view? How does SA Water, or the broader sector, see its place in an AI world?

Amanda Lewry: I see some organisations really seeing what the art of the possible is, and I see other organisations give me five reasons why AI shouldn't be included. I don't think the human mind will give us the best answers to the challenges facing us unless we embrace the power that technology can bring. We can outsource and we already do a whole bunch of the mundane low value add. So all of that can be done for us and we're certainly embracing co-pilot. I did my board papers in four hours instead of two days a couple of months ago just by harnessing AI. We're using it in our infrastructure. We've already got significant use cases in our business. So things like smart networks, if I look at diagnostic, prognostic maintenance, making sure that we understand when assets are likely to fail, sending out work orders and dispatching people to fix them before they fail. We're not quite turning the pipe off automatically, before we can predict a failure.

Ryan Signor: Pierre, I might ask you now of the private sector's view. Is there balance in how the water sector is viewing the opportunity and risk regarding AI? And again, is the ethical thing to do to be trying to harness the power of AI as collectively and wholly as we can?

Pierre Pauliac: The experience shows that artificial intelligence are making less mistakes than human beings. So I would say here is not a risk, it's an improvement. And if it is well-designed with sufficient barriers and checks, you can detect the mistake before it's impacting anybody. Data privacy is maybe the most important. I see it on two levels, the private person's privacy. And for instance, in France, it's a hurdle to implement social tariffs, because water companies, even they consider municipalities, should not know what are the revenues and the number of people in a household. We are trying to overcome it, but actually we are not progressing because of this question of data privacy. And there is security. Let's imagine someone wants to attack us. There are some data like the GIS. If you have twins, a twin can be used to simulate positive actions, but it can be also used also to stimulate negative actions when it comes to AI merits, everybody today is speaking about AI because we have now a sensible relation with AI thanks to GenAI. But GenAI to me, as it is today, we'll see what it will be tomorrow, is productivity. So it's doing things that human beings could do.

Ryan Signor: Can you talk to us a little bit more about some examples in AI for water services?

Pierre Pauliac: I was at the Global Water Summit last week in Paris. There were two topics. Reuse, so recycling wastewater. And the second one, that was not an official topic that everybody was speaking about, is floods prevention. We have software, that is quite cheap by the way, a few hundred thousand euros, that you can install after some simulations in a few months on, let's say, a city, that's improving immediately the protection against floods. Which means that there are a number of events that are creating floods today that will not create floods tomorrow. The same system is helping a municipality to reduce the CapEx that is required to increase the protection against floods because it's making the best of the existing assets and then it's advising, where are the weak points. I'm speaking about flood protection but we can speak about coastal resilience. We can speak about, like we have seawater quality, biological quality, so for swimmers or weather forecasts. So every day we can tell you at 6 am if you should open or close your beach. We do many, many, things like that, that are quite cheap and saving a thousand times what you're investing.

Ryan Signor: The power of AI, Amanda. What's exciting to you? Can you tell us a little bit more about some of what you're driving and where the future may lay?

Amanda Lewry: We're also looking at how do we accelerate our decisions and our planning decisions and optimise those around growth. And we are trialling a whole bunch of GenAI with unstructured data because we haven't got great data sets. They are disparate, but how do get really fast data scraping and decision support in the hands of people, so we can still have some decision support human-centred. But we're also trying to move to online master planning and digital adaptive master planning. So, things like that really excite me. Because I know a dollar spent in planning gives almost a hundred fold based on a change that I might make when I'm operating a plant. So I know that if I can change the way I actually design and build and the plan I have, it has so much more potential to unlock whole-of-life difference for our assets. So that's what really excites me at the moment, is how we can use that information to change the core decisions we're making as we're building brand new communities.

Ryan Signor: Dave, it might be time to bring you in. We've been hearing a lot about applications in the water sector. Are there differences in the uptake and progression of AI across some other sectors? And if there are, who's leading? What can we learn from them?

Dave Mackenzie: I have this slide I present and I say GenAI it’s 70% useful right now, it's 15% unhelpful and 15% wrong. And I think we really obsess about the wrong. Like we really just get hung up on where it's not working and we forget there's 70% of value there. We need to start really looking relentlessly at our back of house processes and how we can drive efficiency there. Microsoft published a great paper with HBR (Harvard Business Review) very recently on the widest use of GitHub, Copilot, and the floor in terms of the efficiency improvement they expect to see is 56 per cent. And I often think we're all from technical organisations, engineering firms and water utilities and others, what could you do with that efficiency when you get it back? And I think that's really the opportunity for AI and I think globally, we're seeing a lot of adoption. It's interesting to reflect on the internet and the calculator. What I learned from that experience is that there's winners and losers and the people who win here will set the agenda for the future and I think it's on us to make sure we're leaning in and setting that agenda.

Ryan Signor: Karlene, again I might start with your perspective. So, what does it mean to you, as a water ambassador for the state, as a community rep and a politician and as a board director, what do you see as the biggest top-down enabler that our business leaders and boards can put in place?

Karlene Maywald: There's a couple of things in this and it's two-sided. The first side of it is, get out of the way and empower the people that know what they're talking about in this space, first and foremost, because most people sitting around board tables have had a career and then bringing to that board table their career that is not engaged with AI in that career. So that's one side. The other side of it is the importance of good governance and ensuring that in the organisation you put in place the mechanisms that enable good governance around any new kind of change management in a business. And I'll use as an example the 70%, 15%, 15% equation that Dave mentioned a moment ago. If we expected every employee to be 100% right every percent of the time, we wouldn't employ them, we wouldn't employ anyone. We actually put in place policies and guidelines and verification processes to ensure that the human mind can minimise its mistakes, because it makes mistakes too. And I would approach AI in the same way. It requires a definite approach that says AI and the 70% is worth pursuing and if we don't get on this train, we're going to be run over by this train. But what we do need to do is actually say, what policies do we have now, and what policies we need to amend, to enable us and empower AI to do good in our business?

Ryan Signor: Amanda, what are the business culture elements that we need to have in place to make these innovations happen? How do we empower people to feel safe to experiment in these spaces?

Amanda Lewry: I think it's got to start with curiosity. I often sit with the team and go, this is the problem we're facing. And I don't have the solutions, but the team does, but also then asking, how could we use AI or how could we use technology to solve this? And they'll give me 13 different reasons how to do it. So I think it's curiosity. But I think it's also courage, because you're not going to get everything right that you swing at in this space, but that shouldn't put you off. And so, ensuring that, if you think it's sort of 60% right, give it a go. You can course correct and move forward. It's very different if you're doing it on a water quality. I get that there are some things that you wanna get a lot more than 60% right, but for most things that we're wanting to swing at with the problems, getting it 60% right and then all the 70, 15, 15 and then improving from there is a lot better than standing by and admiring the problem.

Ryan Signor: Dave and Pierre, I'll ask both of you. Globally, what are some good examples of the things that are in place that have enabled great innovations to happen? What are the key ingredients?

Dave Mackenzie: There's an ISP in the United States that I met with and swapped stories with occasionally and the thing that got AI to move for them, and this is a trend I see across all our clients and various industries, is very strong board support and executive sponsorship. A great example is, this organisation has a mandate to grow, we've got a 20% growth agenda. We will not grow through increasing our head count. The only discussion the executive and the board want to be having is how they're using AI to change how they work and how they going to drive that growth. And I think those kinds of mandates or those line in the sand moments are the things that really rapidly drive movement in this space.

Pierre Pauliac: Climate change is coming. And it's faster than we expected by far, because everybody is saying in 2050 you will have this temperature increase, but it's not that on the 31st of December 2049 you will have the same temperature of today, and on the 1st of January you will gain these three degrees. Actually it's happening now, and what is happening is not only the temperature, it's extreme weather events, much more than before, so more frequent and more extreme. Again, in the last two centuries, when we were far away from the average, it was for a very short period of time. And when we were for a long period of time far from the average, it was not that far. So we were designing on the average. So, no problem, very easy. You take the annual data, you make the average and you have your design. But today we must design on the peaks. And this is making the assets, the need for infrastructure today, almost unbearable because you should triple everything to some extent, even maybe multiply by five or ten, many things. And that's where we will need innovations because we don't have the means. We don't have the means for that and we will be obliged to find solutions, as you said before, that didn't exist because the problems were not the same. And AI will be, to me, should be the first one because that's the one that comes before all the other ones, design and investment in infrastructure.

Ryan Signor: Pierre, Karlene, Amanda, Dave, thanks very much for a great discussion.

+++++

Maria Rampa: I hope you enjoyed this episode of Engineering Reimagined.

As we’ve heard today, the path forward for AI in the water sector will depend on collaboration and a willingness to think differently.

Thanks for joining us on this journey into what’s possible and what’s next for AI in water.

If you enjoyed this episode, hit subscribe on Apple or Spotify and don’t forget to follow Aurecon on your favourite social media platform to stay up to date and join the conversation.

Until next time, thanks for listening.

Unlocking AI's potential in water

Water infrastructure stands at a technological crossroads. While artificial intelligence transforms industries from energy to oil and gas, the water sector remains cautious about embracing the potential of AI. This hesitation isn't unfounded – water systems are inherently local, complex, and critical to community wellbeing, leaving no margin for error. But imagine a future where AI-driven insights transform how we design, build, and operate water infrastructure.

This special episode of Engineering Reimagined was recorded at the Ozwater'25 conference. Aurecon’s Water Industry Market Director, Australia Ryan Signor moderated a panel discussion with guests including South Australian Ambassador for Water and Chair of WaterAid Australia, the Honourable Karlene Maywald, the CEO of Global Water at SUEZ, Pierre Pauliac, SA Water’s General Manager for Growth, Amanda Lewry and Aurecon's Group Director for Digital and AI, Dave Mackenzie.

Together they’ll explore the opportunities and considerations of AI in the water sector and how the public and private sectors are embracing AI now and into the future.

“A dollar spent in planning gives almost a hundred fold based on a change that I might make when I'm operating a plant. So I know that if I can change the way I actually design and build and the plan I have, it has so much more potential to unlock whole of life difference for our assets. So that's what really excites me at the moment, is how we can use that information to change the core decisions we're making as we're building brand new communities.” – Amanda Lewry.

Additional resources



Unfortunately, you are using a web browser that Aurecon does not support.

Please change your browser to one of the options below to improve your experience.

Supported browsers: