Artistic rendition of an underwater scene
AFPL SRA

Episode 2: An ocean of data, with Lyndon Llewellyn from the Australian Institute of Marine Science

In this episode, we chat with Dr Lyndon Llewellyn, Research Manager for the Australian Institute of Marine Science (AIMS), about the past, present, and cyber-physical future of AIMS’s work on the Great Barrier Reef.

Listen and subscribe on Apple Podcasts, iHeartRadio, PocketCasts, Spotify and more. Five star ratings and positive reviews on Apple Podcasts help us get the word out, so please, if you enjoy this episode, please share with others and consider leaving us a rating!

Our guest in this episode is Lyndon Llewellyn, Research Manager for the Australian Institute of Marine Science (AIMS). As you’ll hear in the episode, Lyndon is a multidisciplinary scientist, with expertise in biology, pharmacology, biotechnology and data science. He has a track record of creating and leading multi-organisational projects focusing on biotechnology, marine ecotoxicology, and e-research. He is also well versed in the business of doing science, with a long track record of serving in management roles that have enabled him to gain skills in commercialisation and intellectual property considerations. As his bio on the AIMS website explains, he is passionate about converting science into “real world change” and “maximising the value of marine science to its many stakeholders.”

We invited Lyndon on this podcast because the issues AIMS is dealing with as an organisation are fundamentally connected with larger questions about whether, when, and how we integrate autonomous technologies into our current work practices. AIMS was one of our research partners when we both worked at the ANU School of Cybernetics, and in our work with Lyndon and his colleagues, we were struck by the deep commitment Lyndon and his colleagues have to serving their many stakeholders — including the Great Barrier Reef itself. We also observed that the questions AIMS has been grappling with while considering the use of autonomous technologies for its monitoring programs are not unique to AIMS — and for some, may offer a glimpse of their own organization’s potential future.

With the support of the Erasmus+ Programme of the European Union

This episode was developed in support of the Algorithmic Futures Policy Lab, a collaboration between the Australian National University (ANU) Centre for European Studies, ANU School of Cybernetics, ANU Fenner School of Environment and Society, DIMACS at Rutgers University, and CNRS LAMSADE. The Algorithmic Futures Policy Lab is supported by an Erasmus+ Jean Monnet grant from the European Commission.

Disclaimers

The European Commission support for the Algorithmic Futures Policy Lab does not constitute an endorsement of the contents of the podcast or this webpage, which reflect the views only of the speakers or writers, and the Commission cannot be held responsible for any use which may be made of the information contained therein.

All information we present here is purely for your education and enjoyment and should not be taken as advice specific to your situation.

Episode credits

Podcast Creator – Liz Williams

Hosts – Zena Assaad, Liz Williams

Guest – Lyndon Llewellyn

Producers – Zena Assaad, Liz Williams

Assistant producers – Hannah Feldman, Flynn Shaw

Episode artwork – Zena Assaad

Audio editing – Liz Williams

See episode transcript for links to musical credits and references.

Acknowledgements – A special thanks to Andrew Meares for his helpful feedback on the podcast.

Episode transcript

Liz: Hi everyone, I’m Liz Williams.

Zena: And I’m Zena Assaad.

And this is the Algorithmic Futures podcast. 

[Theme music – Coma-Media from Pixabay]

Liz: Join us, as we talk to technology creators, regulators and dreamers from around the world to learn how complex technologies may shape our environment and societies in the years to come.

Zena: In today’s episode we talk with Lyndon Llewellyn, from the Australian Institute of Marine Science or AIMS. Lyndon has a long record of translating science into real world use, drawing on a diverse technical background in marine biology, analytical biochemistry, molecular pharmacology and data science.

Liz: We’ve invited Lyndon to join us today to talk with us about a technological transformation AIMS is currently undergoing – one in which autonomous technologies begin to help with the massive task of monitoring the health of the Great Barrier Reef. He explains how the health of the great barrier reef is currently monitored, and talks about what scale can mean for marine science. We also explore some of the challenges of scale, particularly around collecting, storing and managing large data sets. Lyndon also shares some of the changes that access to autonomous technologies are beginning to make for AIMS and organisations like it.

Zena: Lyndon grew up in Queensland. The northeast coast of Australia. Which is flanked by the beaches of the Sunshine Coast and the Gold Coast: an expansive coastline that stretches nearly 7000 kilometres. Growing up in this environment really influenced Lyndon’s career choices along his journey.

[Audio clip: Ocean sounds by JuliusH from Pixabay]

Lyndon: So, I’m actually a Queenslander, which I have to say that is being unusual, because in my institute it’s very international. So I’m rare in being a Queenslander within my research institute. But in my research career, I lived in Sydney, I lived in the States, and traveled around a fair bit before coming back to AIMS. And that journey was me shifting disciplines. I started out as a biologist, but then I went and learnt some chemistry and biochemistry. And then went and learnt some pharmacology. And I ended up back at AIMS as a marine pharmacologist, and there’s not many of us. So basically, it’s, if you can imagine a pharmacologist is a person who understands the actions of drugs. So a marine pharmacologist is a person who understands the actions of things like marine toxins and marine natural products, which have an effect upon tissue and organs within animals and humans.

Probably I’ll go back a little bit and say being a Queenslander, my observations is Queenslanders have a very strong sense of ownership of the ocean without even realizing it. Probably more so than any other Australians, because the beach is very much a part of their psyche. I was raised in a small town just outside of Brisbane called Ipswich, which is the western suburbs of Brisbane, like Parramatta is to Sydney. And every Christmas, the whole town would go down to Gold Coast basically, so everyone would pack up, drive down, and go and live in apartments and tent cities down the Gold Coast. And so every holiday, the kids, we all just ended up in the surf, snorkeling, playing in the beach. And I loved it. I just loved being near the ocean. [Clip of ocean sounds in the background by Liz Williams]

And that was a really big influence upon me in that, early on, I just had this desire to be in the water and enjoy being in the water, and then decided, well, see if I can make a career out of it. And that’s how I ended up being a marine scientist.

[Musical interlude by Lesfm from Pixabay]

One of the things that did bring me to AIMS is the fact that it is a multidisciplinary organization. And as I explained about my history, I’m a multidisciplinary person. I’ve transitioned between disciplines. And that’s intentional. I enjoy it. I enjoy the challenge of learning something new. I enjoy the challenge of being able to integrate across disciplines. And it gives me a different perspective on my science. And that’s one of the reasons I joined this organization, because it is not a disciplinary-aligned organization. There are many different disciplines that are required to try and answer the questions that we have to answer and develop the solutions that we need to. And so that I find very challenging, I find quite exciting to be part of.

Liz: Lyndon’s thoughts on the character of the multidisciplinary organisation intrigued us. In our day to day work and in these interviews, we hear the term a lot. But when we dig a bit deeper into what people actually mean by the term, we often find really different ideas about what multidisciplinary work — and multidisciplinary organisations — mean. And so we asked him if we could explore this idea with him.

Lyndon: So to me, multidisciplinary is probably… Maybe I should expand upon that a little bit. It’s multidisciplinary bordering on cross-disciplinary. So, multidisciplinary is just simply multiple disciplines working side-by-side in a simplistic way. Cross-disciplinary is actually when you draw upon the disciplines to come up with integrated science. So it’s the integration which is most interesting. It’s the different perspectives, it’s the different thinking that you will get from different disciplines that improves the quality, and the ingenuity, and the innovation of the science. It’s part of the reason why I enjoy changing disciplines, and one of the things I learned very early on is that while we all speak English, well, English-speaking scientists all speak English, the same word in different disciplines will mean a different thing.

And so when I’m transitioning to a different discipline, as I’ve done a number of times, I have spent time and effort to actually understand the other discipline, not to become an expert, but just so I can have a conversation with the experts. I can use the right words — hopefully in the right way, and I can understand what they’re saying back to me. So we can at least, I can convey the questions I’m trying to answer, and they can convey to me how you might go about it at answering that question.

[Audio: Music by BLACKBOX from Pixabay]

Zena: While Lyndon describes marine scientists through a lens of multiple disciplines and dynamic perspectives, what AIMS researchers do share in common is an interest in tropical marine environments, like the Great Barrier Reef.

AIMS has been monitoring the Great Barrier Reef for over 30 years through their Long Term Monitoring Program. Every year, AIMS sends out a team of experienced divers to observe the health of the reef along specific transects. Divers traverse through the reef, taking photos and notes of the details they observe among the congregation of diverse ecosystems that make up the reef. Divers collect and document this information over time, to observe how details change with our evolving world. 

We’ve been fascinated by these monitoring projects, particularly what their potential impacts are, and how technology might change them in the future, so we asked Lyndon to share the story of the Long Term Monitoring Program.

Lyndon: So the long term monitoring program, plus other monitoring programs that we do have at AIMS, but the long term monitoring program is one of our iconic programs, and it’s been going for quite a number of decades. And it’s always a bit of a surprise for people to find out that its origin was in an employment program. In the early eighties, unemployment was high, government was running employment programs, and it was about the same time that we started seeing, I think it was the second wave of crown-of-thorns starfish outbreaks. And the crown-of-thorns starfish are these quite ravenous coral-eating starfish that occur in nature normally, but in very, very low numbers.

But every now and then, it seems like every 15 to 20 years, there seems to be an explosion in numbers, and of course, they’re coral-eating, so they eat all the coral. But back in the day, some of the scientists of that day, I think it was just quite ingenious in their part was to actually say, “Hey, we can get some funding to actually see what’s going on out there through an employment program for young marine biologists.” And what they did was, they then, they sent these young marine biologists out there to visually observe crown-of-thorns starfish, plus some other important bits of information that might help explain why crown-of-thorns starfish outbreaks occur.

Part of that is, their food. How much coral is there? Is it healthy or not? Are there other indicators that we can see visually that will give us an idea as to the nature of the coral, and how plentiful it is, and how tasty it might be to the crown-of-thorns starfish that would explain why these booms of crown-of-thorns starfish populations occur, and then also why they stop? Not just about what starts them, but what stops them, because they eventually do die out.

So we go out there and we actually document things like how much coral is there? Is it healthy? Is it alive? How much crown-of-thorns starfish are there? Are there other predators of coral? Because there are a few things that do eat coral. There’s another small mollusc that eats coral, which we identify as well. We eventually added observations about iconic fish, fish like coral trout and that sort of stuff, which depend upon the reef for their healthy populations were integrated into the long-term monitoring program. And that’s been going close to 40 years now.

So it’s an ingenious way of starting something up, but it’s also a testament to the organization and those involved who actually have kept it going for that long a period of time, because as you might imagine, monitoring is not the sort of things that gets scientific stakeholders terribly excited until you actually get a couple of decades worth of data. And now, the information that we’re drawing upon, that particular monitoring program, is quite startling.

Liz: Because the Great Barrier Reef is an iconic landmark, there is a strong sense of ownership of the reef among many people. The data that AIMS collects through the Long-Term Monitoring Program has many uses –  including being a source of information for the diverse set of stakeholders associated with the reef and the many people who are interested in the information this data can provide. As Lyndon explains, this is a big part of the challenge AIMS has with this project: They have to be able to scale down the vast dataset they have collected into useful pieces of information for everyone who might ask them about the state of the Reef.

Lyndon: We have a wide array of stakeholders. The Great Barrier Reef is globally renowned, so the global community has a strong sense of ownership of the Great Barrier Reef. Many of the global community will never get to it though. There is the tourism operators, there’s the people who live along the Queensland Coast. There are the tourists themselves. There are the resource managers, so when I say resource manager, it is a marine park and so its protected, and so it needs to be maintained. So obviously, the managers of the marine park have a very strong interest in the data. It is of interest to government, so local councils along the Queensland coastline, they depend upon the reef for much of the tourism that comes in their areas. Then the state governments, Queensland obviously has a vested interest, and the Commonwealth Government has a strong interest in it as well.

So it’s a very, very complex group of stakeholders. And their world views are quite different as well. So if you can imagine a local city council and the people who live in a particular town or particular tourism operator will have a very strong sense of ownership of a reef. And the operator may have a very strong sense of ownership of their patch of the reef. Whereas the global community, all they know about is the Great Barrier Reef.

So we get very complex questions. They’re simple questions, but they’re complex, because we will get questions from a tourism operator who may say, “Well, how is my patch of the reef going?” And then we’ll get questions, as we often do, from the global community and from the governments who will say, “How is the Great Barrier Reef doing?” And we can’t answer those questions with a 100 page report. We need to be able to try and distill that down and get it across to people in the simplest, but the most factual way we can. But we need to be able to scale our answers. We need to be able to do monitoring at scale in terms of looking across the Great Barrier Reef, but in a way that can be downscaled to where we can give answers to people in different locations and do so in a way that maintains their confidence.

And part of the scaling down issue is that when you do something at scale, you have a lot more data. And so, of course, when you downscale something, you lose data. So how do you do that in a way, but maintain the power of your analysis?

Zena: Let’s take a moment to picture the challenge scale presents for AIMS here. The Great Barrier Reef is the World’s largest coral reef system, covering an area a little larger than the size of Italy. And this rich ecosystem is one where, depending on your question, you might need to know something about very small scales — for example, the properties of the tiny photosynthetic algae called zooxanthellae that live in symbiosis with corals and give them their colour. On the other end of the spectrum, you may need information on a larger scale, like how has the amount of coral in the Reef changed from one year to the next.

Liz: Distinguishing useful information –  be it at a small or large scale – amongst a vast collection of data is not a simple task. And, if one of the major goals of the Long Term Monitoring Program is to monitor the health of the Reef over time, we wanted to understand what the observers in the Program are looking for. So we asked Lyndon: What does a healthy reef look like?

Lyndon: So that’s a really interesting question, what does a healthy reef look like? And it really does depend upon your perspective and what your experience is. So if you’re someone who’s only ever seen a David Attenborough special that only ever shows the best possible condition of reef, that’s your perspective of what a healthy reef would look like. However, if you come from a location where the reefs have been already heavily fished, there’s already pollution going on, but they’re still productive, that might be what your perspective of a healthy reef is.

So it really does, it does depend upon your perspective. However, we tend to take the view that a healthy reef is one which provides a lot of services to humanity, or is able to function as well as possible to provide as much of the services that they normally provide. Now, that’s an interesting concept to some people in terms of ecosystem services. And one of the easiest way maybe to explain an ecosystem service is that when you eat a fish at night time, that’s been produced by an ecosystem. It’s been produced by an ecosystem that’s been cycling nutrients, that’s been producing food, that has enabled that fish to breed, to grow, and get to a point where it can be fished, and then get to a table.

So it’s actually a service that the ecosystem has provided you in enabling that fish to be, and to be in a condition that you would want to eat it. There’s a very wide array of services. One of those services is aesthetics, how pleasing to the eye can a reef be? And so we tend to take the view that a healthy reef is one that produces as many ecosystem services as it possibly can depending upon the situation that it is in. And I add that little bit there, because it will be different in winter, as it will be in summer. They’re the sorts of things that will change through time.

And so time is really quite important at the scale at which you want to look at something. It might be slightly different in summer, it might be slightly different to several summers ago. Each summer will be slightly different. And that gets onto one of the points that’s been an issue of late in that people would have heard about, coral bleaching. And coral bleaching is a phenomenon that’s been around for some time, but what is different in recent years is this phenomenon called mass coral bleaching.

[Musical interlude by Bluemount_Score from Pixabay]

So, corals will bleach, but what we’re seeing in recent years are these mass coral bleaching events where many corals in many reefs are bleaching at the same time. The bleaching is a process where the coral expels its symbiotic algae. So corals are an interesting organism in that they aren’t just an animal, they host an algae called zooxanthellae. And that’s what gives it its color, is the zooxanthellae. And they depend upon those zooxanthellae because they photosynthesize and produce nutrients that allow the coral to live.

The corals produce some of their nutrients from predating plankton and things like that, but the bulk of their nutrition comes from these zooxanthellic algae. But corals tend to live at their limits. That’s one of the ways that animals occupy different niches and ecosystems, and so corals actually live near the limits of their temperature tolerance. If it gets too hot too quickly and stays too hot for too long, so it’s not just about being hot, it’s about how long it’s hot for that eventually the response of the coral is to expel the algae. And there’s no real firm answers as to why it gets rid of the algae. There’s some suspicion that they start producing things like free radicals or things like that. And so the coral will say, “Look, I want your nutrients, but I can’t take those sorts of things as well.”

And they’ll expel this symbiotic algae, and when that happens, it reveals their coral skeleton, which is white. And the coral itself have got no color. And so suddenly, it looks bleached, it looks white, even though it’s actually got the coral tissue over it. The symbiotic algae can survive for some time in the water, and if the conditions get better, then the corals can grab those symbiotic algae and bring them back in and get back to where they were. So the bleaching isn’t a death sentence, it’s a bad period for them, if the temperature disappears and it cools down they can recapture the algae and head back to where they were. It’ll take some time for them to recover, just like any animal gets sick.

So, as you might imagine, when that happens, then the ecosystem services that those coral reefs provide are nowhere near what they could be. And so in that case, obviously, it’s an unhealthy reef. And of course, if the corals die, well, it’s more than unhealthy, it’s not even a reef, it’s now simply dead coral. So, that’s one of the reasons why we monitor these things to actually look at how the coral reefs are changing through time. And in addition to that, we monitor a lot of other information like seawater temperature, and salinity, and those sorts of things. How murky the sea water is? Because it is those things which can drive or compound each other that affect the health of the corals that are bathed within that seawater.

So we look at the seawater temperature. We can look at what the forecasts are from summer, and we can start thinking, “There might be a tough season coming up, and so there might be some bleaching may occur. We need to be ready to think about whether or not we need to do anything.”

[Musical interlude by Coma-Media from Pixabay]

So one of the really key issues in the coral reefs and Great Barrier reefs is this concept of cumulative impacts. And that’s why it’s often really quite difficult sometimes to give a simple answer to many questions. And it’s also part of the reason why it takes so many years for monitoring programs to produce data that allows better understanding, because you have both natural processes, acute processes, and of course, processes that are driven by human activity. So you need to actually understand natural variability, and then understand, is the variation occurring unnatural, or not?

And in a system as big as the Great Barrier Reef it takes a while for you to actually document that natural variability. And I suspect even as we continue to monitor, we’ll start capturing other variable processes, which operate on decadal scales, rather than even monthly or annual scales. And so, cumulative impacts is a really critical issue, because when people talk about the stressors and impacts on reefs, you often focus on a single impact.

So you can imagine if I’m sick, and I get another disease, then obviously my ability to deal with this second disease is hampered by the fact that I’ve already got another disease. So most people can get that. The same thing is true of an ecosystem. So you can imagine if a coral reef, it’s just been hit by a cyclone. The cyclone has damaged the reef, it’s driven away all the fish that live on the reef. It’s driven all the critters that live on the reef down into the inner structures of the reef, and so they’re not there to clean and do the stuff that they do.

And then it gets really hot. The ability of that reef to deal with being really hot, is different from the year before when there was no cyclone. And that’s where this concept of cumulative impacts and trying to work out how the reef might respond to different stressors is quite difficult, because you imagine cyclone, crown-of-thorns starfish, really hot summer. Not good. And part of the reason that you’ll hear that there’s management programs for things like crown-of-thorns starfish is that something that people can do something about. They can go out and they can harvest them and remove them. That reduces one of the stressors, which means that the reef doesn’t have to deal with that, and just has to deal with other stressors.

[Musical interlude by Lesfm from Pixabay.]

Zena: How the reef changes with time is influenced by many factors, including cumulative stressors, such as natural disasters and the increased temperatures that come with climate change. When thinking about these stressors, it’s important to remember that the reef itself is not one thing: it’s a system of many different organisms and structures living in symbiosis. For example, there are about 600 different kinds of coral in the Great Barrier Reef. As Lyndon points out, each of these types of corals can respond differently to these stressors over time.

Lyndon: Different corals respond in different ways. When we talk about coral, there are hundreds of species. Some of them are fragile, some of them are fast growing. Some of them are these big, massive brain corals that people will see. They’re centuries old. Sometimes over a 1000 years old. So, every coral is different, so it’s hard to generalize and say how corals will respond. But what we’re seeing is that in some reefs, those reefs are recovering faster than we thought they might from cyclones. But one of the things that we’re starting to burrow into is that that recovery may be from the faster growing corals.

So in some areas, you will get faster growing corals, just like in your garden. If someone comes in and rips up all your plants, the weeds are the first things that grow back. And so it’ll look green, but it’ll be weeds. We will see coral recovery in some areas which is faster than we may have thought, but it’s the type of corals that are growing back that’s important. Now that might be a natural succession process where the weedy corals come in first, and eventually the stronger, more robust corals will take over.

But we need to be able to make sure that there’s enough time for that process to occur. So we want to try and reduce the opportunity or reduce the chance that there might be a bleaching this year, and then there’s a relatively quick recovery, but the nature of the… it might be a poor quality recovery as opposed to a good quality recovery. So if it then gets hit again, you might just end up in a process of continually staying in an area of poor recovery, rather than long term, good recovery. I’m hoping I explained that one particularly well. It’s a bit complex, but if you might imagine a farmer who’s continually growing a crop of weeds. They want to be able to grow their crops. And that’s what we need to do on the reef, we need to give enough space for them to get past that early succession phase and get the really good corals growing.

[Musical interlude by ZakharValaha from Pixabay]

Liz: Experienced divers manually collect and collate data about the reef, to monitor changes over time. One of the challenges of collecting data manually is the time it takes to do this job. The reef stretches over 344 thousand square kilometers – so a sea area larger than the size of Italy — making it very challenging to monitor every portion of the reef over a consistent time frame.

Large scale environmental monitoring like this is increasingly being done with the help of autonomous technologies — technologies with sensing capabilities that can either be distributed throughout or move across vast swathes of land and sea to collect and send back data about their surrounding environment. At AIMS, they have been considering adopting some of these ideas for their monitoring programs.

Lyndon: This goes to a really, really simple proposition, which is one of the things that we learnt from the defense force, which is the concept of force multiplication. We only have so many scientists. This is an issue the whole of Australia faces. We’ve got a massive amount of ocean out there, and maybe a few thousand marine scientists. And so they will never, we just don’t have the people to be able to learn and understand our ocean territory.

So you either need to get enough marine scientists that can do that. That’s not going to happen. Or what you do is, you allow the scientists to be able to do more with the available time. And the obvious solution there is technology. We need to be able to double, treble, quadruple the amount of information that can be produced by our scientists for the amount of effort they put into it.

So if you use technology, then what you can do is at the same time someone might be diving, or snorkeling, or underwater gathering data, at the same time you can have other technologies out there capturing data at the exact same time. And capturing the contextual data. It might be capturing the data around the scientists as they’re doing it. Or they may be getting the exact same data, just over a bigger area. So this is really important for us to be able to get, to increase the amount of data and information that we capture about the marine systems for the same amount of human effort.

There are some obvious challenges there. No technology as yet has enough energy or endurance that we can get away from using a ship. So we’re monitoring quite remote locations. Our divers can’t swim there, nor can any technology. And we need a ship. And we need the ship to be able to get our divers out there, and the technologies. And so it gets quite complex, as you might imagine, running a mission that has both people in the water, and devices in the water. So that’s one of the challenges is, not only just getting the technology that can compliment the human in the ocean, but that you can actually manage it in a way that what comes back is meaningful.

It’s very easy to produce numbers, it’s more difficult to convert those numbers into data. It’s even more difficult to turn that data into information. And that all takes a fair bit of effort. And if you’re producing an enormous amount of numbers, that makes it even more challenging in terms of your workflows and the processes you have in place to make sure that you maximize the conversion of those numbers into useful information. So look, it all boils down to force multiplication. We just need to get much more information, quality information for the amount of human effort that goes on.

And probably, the other aspect to it, and this may not be as important, but I think it’s quite important, we need to reduce the amount of time that our scientists then spend being the muscles to capture the data. We need them to spend their time and effort applying their brains to answering the questions and analyzing the data. We need to them spend less time being the machines, and actually capturing the data. We don’t want that to completely stop, because the machine won’t develop a gut feel.

There’s a lot to be said about someone who’s a trained marine scientist, he’s been observing something for 20 years. We need to have that, and we need to continue to have them exposed to it to see things because obviously there’ll be things happening in the subconscious. Things will get processed without you even knowing it, and so you’ll have a perception about things, which only comes from seeing things and sensing things. So we don’t want to completely remove people from the water, we just need to be able to do much more for the amount of human effort that gets put into it.

Zena: The use of more autonomous technologies for data collection could bring more data, faster, about the state of the reef, but this choice has a possible downside: it reduces the time divers spend in the water manually monitoring the reef. Divers bring with them deep experience and knowledge of the reef. This thinking led us to questions around what the balance would look like between maintaining the benefits of having a human collect the data vs the benefits of having data collected through autonomous technology capabilities?

Lyndon: So data captured by a machine will be completely unbiased. It will be completely unemotional. It will have no world view to it. And that’s got a real plus to it. There’s some negatives to human biases influencing how you do things. But because it has, in that case, every mission to that machine will be a new one. So every mission it’ll learn from scratch. You know what I mean? So it won’t go, “I was at this reef last year, and there was something going on.” It won’t have those memories. It won’t have that basically to draw upon, but also to provide a context for what they’re observing. And I think that’s probably the biggest thing, is the memory of what’s out there. What’s that perspective? What’s the context into which the observations, at the moment, can fit into?

One of the techniques that we use, it sounds funny is, we’re actually just simply towing very experienced people along behind a boat, and they’re making observations. They can react. They can actually see something and go, “That looks weird.” And they can go down and have a look at it. And they can say it looks weird, because of the years of experience and the context they’ve got. It’ll be a long time before a machine can do that. So I think that’s a really key thing, is that every data point, I think, that a machine gathers is a data point, whereas with the humans, every data point is an observation that they then embed in their recollections, and memories, and understanding of the system.

And another difference between the human and the machine-gathered information is potentially experimental design. So with a human, and I’ll use a very simple experiment, or way of doing things where people put out transects. They just simply lay out a tape measure and just make observations along a tape measure. That might be difficult for a machine to do, because the ocean is dynamic and it moves. So what you might do is, you simply might make the machine fly around a huge area and hopefully capture the area of interest. And so the experimental design will be different between those two things.

So I think it’s going to be an interesting one, and I don’t have a really simple answer for you on that one because the standard way of comparing result, of seeing how good something is, is to compare the new with the old way of doing things. And the assumption there is that the old way of doing things is correct. And so a human scientist may have said, “I think the number is 20%.” And you might have 20 people who all say, “Yeah, I think it’s 20%.” Or, “20 plus or minus some percent.” And then the machine might come along and say, “I think it’s 40%.” Is that because the human has made an inherent error in how they’re doing things? Is the machine right? Not knowing who is right is going to be a real challenge for us.

And I have no answer for that one, because that’s going to be the real crux of things is because we have a huge amount of history in this space, how will we be confident in that the new way of measuring things is as good as how we’ve done it for many decades? And that then flows to how will our stakeholders be confident in our confidence that the numbers are correct?

Liz: As Lyndon spoke about the differences that come from different methods of data collection, whether they are collected manually or collected using autonomous technology capabilities, we started to wonder about how AIMS is thinking about transitioning to these new methods. How will they begin to develop confidence in the new methods they’re introducing?  

I suspect that there’ll be a transition process where the machine-gathered information will be considered a subset of the human-gathered information. And so, I’ll try and explain that with an example. Many years ago, I looked at some data that was gathered by an AUV in the Great Barrier Reef, looking for sea grass beds. The challenge there is that when you look at something under water, it’s green. It could be algae, or it could be sea grass.

Now, if I was in the water, and I couldn’t have looked at this because it’s too deep, so I can’t go there. But the intent was to actually look at sea grass, but all we could actually do was say that there was sub aquatic vegetation there. But that is still useful information, it’s just not what the original thing was. If it was in shallow enough water, then a person would’ve gone along down to say sea grass/algae, sea grass/algae because they know enough. But because we were looking at an image back in the laboratory, you can’t go back and check. You can’t go there and look at things.

So that, I think, will be the transition where there will be, potentially, two types of data at different resolutions, or different taxonomic resolutions, or different spatial resolutions. And hopefully, the monitoring will be designed in such a way that they will help answer the same question. But probably, it’ll be baby steps, I think, and that will give people who are advocates of technologies a chance to actually prove that technology works, but also be a transition process for those who are cautious about the adoption of technologies to actually, “Yep, okay. I’m starting to get some confidence in the information that’s being generated.”

But that process, in terms of quality control, when we look at our existing monitoring program, we put a lot of effort into observer checking, in training, and all those sorts of things, which are the sorts of things that you can’t do with a machine yet. We can train a machine through machine learning and AI, but it’s not the same as having people who are experts train a human being in what a particular coral species might be, or a particular sea grass might be. But we’ve got to be a little bit, I think, innovative in how we transfer knowledge to the technologies, and how we integrate the information that comes back from them into the broader information ecosystem, for want of a better phrase.

Liz: The integration of autonomous technologies will represent a major change in the ways AIMS achieves its mission of supporting both the protection and sustainable use of our marine heritage. We asked Lyndon to share his thoughts on the ways this change might impact AIMS as an organisation.

Lyndon: Well, it’ll be a fundamental change. It’ll change how we plan. It’ll change how we do. And what I mean by that is, we’ll need to have technologies ready to go. Our ships will look different, and how we do things on those ships will be different. We’ll have people in the water at the same time as technologies. How will we manage that? How will we manage that so it’s effective and safe? We need to make sure that divers don’t end up in any sort of perilous situation. So we need to make sure they’re okay. And we need to make sure that the devices in the water don’t pose any challenges to them. We’ll need to have technologies in place to rescue the technologies. We have technologies to rescue people, we’ll have to have technologies to rescue the technologies in case there’s a problem.

We’ll need to have control systems on our ships. We’ll need to have an extraordinary amount of data storage and an extraordinary amount of data analysis, because the biggest thing that’ll happen here will be a massive data explosion. We’re already starting to see this. We’re seeing the amount of data that can be produced on some of our field trips has just ballooned. What do we do with it all? The inherent nature of scientists is to not throw away any bit of data, but is there something that we’ll have to do where we’ll actually say, “Look, we just can’t keep it all.”

But does that mean that we’re sacrificing the potential to answer a question in 20 year’s time that hadn’t been thought of? And that’s a big debate that’ll happen amongst a whole raft of people. We’re seeing now a lot of value in photographs taken a 100 years ago. Well, think about in a 100 year’s time, there’s minimal photographs now, so think about how much value those photographs will have to people in a 100 year’s time, looking back on how things were.

So it’s going to be really interesting, it will have a really fundamental impact upon our operations, our planning, how we think about things, how we analyze it. And then the final step is, how we communicate it. How do we take all that information? We struggle now with the amount of data to synthesize it and put it into a form that our different stakeholders can consume in a simple and straightforward way. Well, when we’re getting 10 times or a 100 times as much more data and information, it’s going to be an even more difficult synthesizing process.

Zena: AIMS is not the only organisation grappling with this kind of data problem. Autonomous technologies deployed at scale are leading to an explosion in the amounts and types of data we collect every day. This data has to be stored, processed, and be turned into useful, understandable, and — most importantly — more concise information.

Lyndon: I think there’s been a really fundamental shift in the data world, which might sound a bit strange. I think many people will be familiar with the concept of metadata, because of phone records and we’re only keeping the metadata, not the data, so it’s not actually the data. There’s been a really fundamental shift in data where in the past, the metadata would refer to a large data record, and I’ll use the phone as an example. It’ll be such-and-such called such-and-such at this date and time, and blah, blah, blah. But the conversation might be an hour-long recording.

What’s happening now is it has been flipped around and that basically we are now able and what is happening is that every data point will have metadata, and almost an infinite amount of metadata can be attached to every data point. So that’s a really different thing when you’ve got a data set of a 100,000 data points, but one metadata record, and now you may have a 100,000 data points, each and every one will have a metadata record. And we can actually go back and add metadata if we learnt something about those data points. That’s a really, really powerful thing then, because we can actually tease data sets apart, and we can reassemble them if we want to.

So in theory, you can take a data set where every data point has a metadata record, mix it all up, get something really interesting, and then pull your original data set back out. So we can do integrations we never thought about before. And that is starting to happen. But as you might imagine, it’s a really fundamental change in how many people are thinking about data, because it’s got some possibilities and potentials that you don’t realize until you start doing that mix and matching.

Liz: Marine science is one of many disciplines currently grappling with the challenges of exponential increases in data collection. So who is AIMS learning from in this space?

Lyndon: Anyone that can help. You’re right, it’s a massive challenge which is being confronted by many, many people. And we are talking to other organizations in the field, but we also need to look beyond marine science, because many solutions will come from elsewhere. Because while we might think some of our challenges are unique, they’re not always. And maybe I’ll use this as an example. A lot of the data that we’re producing now is video. And that’s very high bandwidth. And yet, somehow these streaming services, they have algorithms able to compress and transmit really high bandwidth stuff really easily over simple networks. That’s all commercial, in confidence, commercially valuable. And I think one of the challenges that we’re going to face as some of our solutions are in behind commercial borders. But could you imagine the boon to us might be is, if we can get access to an algorithm like that that can compress a video of one of our field trips from a gigabyte down to a couple of megabytes, that would be a huge benefit to us.

Because we can now get so much data, because in theory, you can simply twist a knob and just simply increase the number of photographs you take, or you suddenly get video versus a photograph. In a lot of cases, there’s no change, right? So a lot of the data that we’re capturing now may be completely superfluous. It may actually add no information. So, part of what we need to do is actually understand how much information is in a scene or a photograph, and try and find some way where we identify those images which have got some information to them.

And so, one of our techniques, we put videos under water and we look at what the fish around them are, and we document what those fish are. And many minutes of those video, there’s nothing, right? Well, I shouldn’t say nothing, there’s the background, there’s the reef, which doesn’t change, right? So basically, it’s a lot of video of a photo. So can we take all those images that look the same and basically go, “Yeah, no, you’re the same.” And we take 20,000 frames and make them into one. And we’ve gotten rid of 19,999 frames that actually showed no more information than that one frame. But then there’s 500 frames, which have information. So things like that are going to be really important for us. How do we actually winnow out what’s got valuable information, or compress those sorts of things into what is the base amount of information so we’re not storing, for decades, images of the same scene that doesn’t change for hours on end.

[Musical interlude by Coma-Media from Pixabay]

Liz: Because the Great Barrier Reef is such a large and expansive ecosystem, the scale of the data needed to understand the health of the reef can be, as Lyndon described, overwhelming. These enormous data sets that AIMS collects have far reaching impacts that extend beyond the needs of scientists working directly with the reef. As Lyndon sees it, one of the important things AIMS can do is help everyone — not just scientists — connect with the Reef.

Lyndon: I think one of the key things of the data that AIMS collects is it brings what is often out of sight and out of mind, into sight and into mind. We deal with a system, or systems, that are so big it’s difficult for people to comprehend. It’s difficult for us to comprehend. And we’re dealing with systems that are well away from anyone. And so, sometimes the only people that get to see them are our scientists. So a really important impact of our science is it actually brings it to them, it brings it into it, it makes it real for them as well. And I think that’s really important from a number of angles. It makes it real for the community, because they actually know it’s out there, they know what condition it’s in. And they can then have a sense of comfort or that something that they have a sense of ownership of is it in a good state or not? But then you’ll have a more practical aspect where the managers will actually have a far more detailed view of what’s going on. They’ll actually have a more numerical view, for want of a better word, the community might have an aesthetic view of what’s going on out there, whereas the managers might have a more functional, a practical perspective and just how are things going and do I need to make a management decision? Do I need to do something? As opposed to the general community who just, they just want to know how it’s going out there and what it might be like in the future. But I think that’s the really key thing is just bringing into people’s world something that’s a long way from them.

Zena: For AIMS, scale comes with very large data sets that are collected and analysed over much shorter time frames compared to manual data collection and analysis. As monitoring of the great barrier reef is conducted as a function of time, scaling this process may directly benefit our understanding of the Reef as a system.

Lyndon: So the example I’ll use about the benefits of scale is tides. So you can imagine if you put a stick on the beach and you went down and had a look at that stick once a day, you would see the water’s in a different spot. And it would take you a really long time to work out that there was this phenomenon called tides. And then you’d maybe put a few more sticks on the beach in different places and then get a feel for the fact that the tides are actually different in different locations.

Compare that to now where we have tide gauges everywhere that collect data every second, not every day, that gives us an incredibly different perspective of tides where now we predict what the tides are going to be anywhere in the world at any time and we can forecast some decades ahead, because we understand the process. We were able to understand the process what drives the tides, and then look at the observations that allowed us to validate that.

So, being able to get observations every second versus every day or every week, in that tidal example, gives you a completely different perspective, and being able to look at it in different locations gives you a different perspective. So that’s the value of being able to upscale in terms of spatial scale and temporal scale. And then you have the next thing which is then, how do you then reduce that back down to something which is actually useful and comprehendible to someone?

Now there are, obviously, prime locations up and down the reef, and so we engage very closely with tourism operators. We engage with them to give them some idea. There are many inter-organizational groups where the different experts and different perspectives sit there along with stakeholders, and they will talk with them about is there a risk of bleaching that might happen this year, or are there things that they might be able to put in place that’ll allow them to deal with some localized effects. We can downscale some of the data we have now to be able to say that the risk in one area might be higher than another area that allows that tourism operator or that area to try and put in some area, or at least be a bit more alert to what’s going on around them.

And probably, a really practical example of that is with the crown-of-thorns starfish.

Zena: Crown-of-thorns starfish are those coral-eating starfish that occur in nature –but normally in very low numbers..

Lyndon: People can help identify whether or not the crown-of-thorns starfish, the wave of population boom is approaching an area, then you can target the removal of those crown-of-thorns starfish to those tourism operators in those areas. So you can do an immediate response and we can look after those locations and give them a higher priority.

[Musical interlude by Alex_MakeMusic from Pixabay]

Liz: One of the important stakeholder groups AIMS works with on the Reef are the Traditional Owners – Aboriginal and Torres Strait Islander communities who have been,and continue to be,actively involved in looking after the Reef for the last 60,000 years. We asked Lyndon whether there was a connection between the data they are planning to collect and the work AIMS as an organisation has been doing with the Indigenous community.

Lyndon: So we are putting a lot of effort into our indigenous engagement, and we’re trying to do it in a better way than we feel it’s been done in other places or in other times. We’re trying to do it in a way that we recognize that the indigenous communities are scientists in their own rights and have been as well. So often, some of the approach in the past had just been about give us information that’ll make our science better. Our approach is that the melding of the two approaches can be obviously far better than operating independently of each other. And that goes back to a bit to my earlier comments about the changing in disciplines. You need to put a lot of time and effort into actually understanding the words and thinking of the other discipline. Well, the same goes there where you need to think about the words and thinking that are in behind indigenous science.

And I use this example with others in that observing science in our world is putting a device out there that then sends numbers that we get collecting the database. In indigenous science, those observations have been going in the reef for centuries. And that data is stored in their storytelling, and their songs, and things like that. That’s what their databases are, they’re just not on the computer. And that’s just a really fundamental different way of thinking about things. And so you need to build up trust between the disciplines, between the Western scientist, for want of a better word, and the indigenous scientist, to actually have that exchange of knowledge between the two disciplines, and to actually have a language that the two disciplines can understand.

And there’s also got to be mutual respect. There’s got to be respect for the fact that how science have been done in the indigenous communities. That way, of course, it gets closer and you just end up with a better understanding of what’s going on out there, and a better understanding of what’s been going on out there, much longer than our observations have gone back.

[Musical interlude by Coma-Media from Pixabay]

Zena: And finally, we asked our guest what socially responsible algorithms mean to them.

Lyndon: The socially responsible algorithm is often associated with artificial intelligence, machine learning, and almost, I suppose, removing humanity from the loop. The non-human-in-the-loop algorithm. But the reality is, so much nowadays is already algorithm-based. We have human-in-the-loop algorithms. I mean, much of our science, we run algorithms and codes and some are just extraordinarily complex. And they produce numbers that we believe, and we expect others to believe as well. And then people will make decisions based upon those numbers.

So, what is a socially responsible algorithm? I initially went to an algorithm that actually is a true portrayal of what you’re looking at, the old WYSIWYG concept I suppose, that it doesn’t mutate or doesn’t transform what you’re looking at into something completely different. But that’s probably too simplistic, because some of the answers will be quite, they’d be quite other-worldly I suppose, for instance, all the physicists are talking about different dimensions and stuff like that. I really struggled with this one, I got to say. What is a socially… And honestly, I only got to the point where it was like, well, we’re already doing it, and I don’t know how responsible it is what we do. You know what I mean? I don’t know if it fits in with the understanding. And I think it’s going to be a real challenge when you completely remove humanity from that process. Does it lose responsibility by being dispassionate and having no human involvement? I don’t know? I’ve struggled with that one.

So we do a lot of stuff with code.  I always explain myself as a copy and paste coder, not a professional coder. But it’s a really wonderful change in the world where we’re seeing now open data is becoming the norm, not the exception, but even more so is the open code. You’ve got these brilliant people all writing code that do amazing things with the numbers that you have, and then they just give their code away. It’s just amazing. It’s just incredible altruism, right? So it’s amazing stuff. But in doing that, that code can transform numbers into something quite different from the original thing. Is it a true representation of what I’m trying to convey? Of what I’m trying to explain? Because you can transform, you can get from one point to another with an infinite number of ways of writing the same code.

And so, I can produce really clever visualizations. I can produce really clever outputs with a piece of code. Is it right? I don’t know. It looks good, but is it right? So maybe that’s what I might define as social responsible algorithm is that it’s something that converts numbers into something else. And what it converts it into is right. But that rightness can be both a social thing and a mathematical. The answer might be four, but there might be a social perspective there where the rightness of the answer is not just about what the number is.

Liz: Thank you for joining us today on the Algorithmic Futures podcast. To learn more about the podcast, the Social Responsibility of Algorithms workshop series and our guests you can visit our website algorithmicfutures.org. And if you’ve enjoyed this episode, please like and share with others.

Now, to end with a couple of disclaimers.

All information we present here is for your education and enjoyment and should not be taken as advice specific to your situation.

This podcast episode was created in support of the Algorithmic Futures Policy Lab – a collaboration between the Australian National University School of Cybernetics, ANU Fenner School of Environment and Society, ANU Centre for European Studies, CNRS LAMSADE and DIMACS at Rutgers University. The Algorithmic Futures Policy Lab receives the support of the Erasmus+ Programme of the European Union. The European Commission support for the Algorithmic Futures Policy Lab does not constitute an endorsement of this podcast episode’s contents, which reflect the views only of the speakers, and the Commission cannot be held responsible for any use which may be made of the information contained therein.

[Post-disclaimer audio: Ocean sounds, by Liz Williams]