Apple Podcasts | Spotify | YouTube | RSS | Android
With 25% of marine biodiversity found on coral reefs, it’s vital they are kept healthy! On Climate Curious today we hear from marine biologist, data scientist, and TEDxLondon speaker, Ben Williams, on how he’s utilising AI models to scan through years of underwater audio. His work focuses on measuring the health of coral reefs in an effort to help keep them ship-shape, as well as keeping an ear out for illegal bomb-fishing!
Listen to Ben’s TEDxLondon talk here
Created by TEDxLondon
Executive produced by Josie Colter
Produced by Ben Beheshty
Curated by Deesha Chandra
Hosted by Maryam Pasha and Ben Hurst
Communications by Tara Cooper and Issey Gladston
00:00 Introduction
01:45 Ben Williams
03:50 Understanding coral reefs and climate change
06:13 The importance of coral reefs
09:56 Exploring coral reef soundscapes
13:18 AI in coral reef conservation
16:46 Coral reef restoration techniques
17:36 Introduction to AI in coral reef restoration
18:03 AI applications and challenges
22:39 Community involvement in coral reef projects
24:46 The issue of bomb fishing
30:55 Future directions and innovations
34:30 Conclusion
Transcript
[00:00:00] Introduction
[00:00:00] Ben Williams: This began quite simply by kind of chucking the sounds of healthy reefs at an AI model alongside those from degraded reefs. And we found that this model could actually do pretty good job at identifying the difference between these. So if we gave it new recordings from a reef it hadn’t heard before, it could tell whether this came from a healthy or degraded reef.
[00:00:21] Ben Hurst: This is climate Curious, the podcast for people who are bored, scared. All confused by climate change.
[00:00:28] Maryam Pasha: I’m Maryam Pasha, the director and [00:00:30] curator at Telex London, and the co-host of this podcast alongside the amazing Ben. Hi,
[00:00:34] Ben Hurst: I’m Ben Hurst, activist and advocate exploring what positive masculinities can look like and self-confessed climate normie.
[00:00:45] Maryam Pasha: Ben, you know what we get to do today?
[00:00:47] Ben Hurst: Do I know what we get to do today? I feel like I have a good idea, but are you gonna tell me? I’m gonna say
[00:00:51] Maryam Pasha: we get to do something. That is one of my favourite things to do. Yeah. You’re like, what is it?
[00:00:55] Ben Hurst: Yeah. What is it? Is it garden? Is it what are talking about? No, you, we have
[00:00:59] Maryam Pasha: to come back to your [00:01:00] garden.
[00:01:00] Maryam Pasha: Yeah, yeah. We’ll come, we come back to your garden. Um, it’s so we get to actually. Talk to one of our telex London speakers. Ooh. And the reason exciting, the reason this makes me happy ’cause you don’t actually know why this makes me happy. You’re just being generous. Yeah, yeah, yeah, yeah. It’s not that same.
[00:01:14] Maryam Pasha: ’cause when you all get to listen to one of our, one of the amazing speakers at Telex London, it’s like the 12 minute highlights. Mm-hmm. You know, it’s, it’s. Um,
[00:01:24] Ben Hurst: refined. Yeah. Refined. Yeah. Yeah. Yeah.
[00:01:26] Maryam Pasha: But behind the scenes, there’s so much conversation and so much, [00:01:30] much depth of knowledge and so many things that we just don’t get to talk about.
[00:01:34] Ben Hurst: Right.
[00:01:35] Maryam Pasha: In what? On that, on stage. And so we get to do that here. So that’s why I’m excited. We’re
[00:01:38] Ben Hurst: gonna unpack it. We’re gonna explore it. Get into the, the nuts and bolts of it. Yeah. The meat and potatoes.
[00:01:44] Maryam Pasha: So.
[00:01:45] Ben Williams
[00:01:45] Maryam Pasha: We are joined today by the wonderful Ben Williams. Uh, he is a marine biologist and data scientist and is completing a PhD at the University College London, and the Zoological Society of London, which I always have to pause and say properly.
[00:02:01] Maryam Pasha: Zoological
[00:02:01] Ben Hurst: is a different word. Zoological. Yeah. Yeah, yeah.
[00:02:03] Maryam Pasha: Um, he’s on a mission to protect our natural world using the power of ai, which we all know because we’ve all watched his talk.
[00:02:10] Ben Hurst: Mm-hmm.
[00:02:11] Maryam Pasha: Right.
[00:02:12] Ben Hurst: Yeah, I was there. Yeah. I mean, I was there. Yeah, yeah, yeah. I was there. Remember,
[00:02:16] Maryam Pasha: so we’re gonna get into it, but broadly, Ben’s work focuses on using AI to study the soundscape of coral reefs to support conservation and restoration.
[00:02:24] Maryam Pasha: And he collaborates on one of the world’s largest reef restoration [00:02:30] programs, which is led by Mars Sheba. It is called the Mars Shiba Hope Rose campaign. And that are, those are the chocolate bar and cat food. Oh, is
[00:02:40] Ben Hurst: it actually,
[00:02:41] Maryam Pasha: dudes? Yes. And you also works with Google Deep Mind, which is very cool. We’ll really get into all of this.
[00:02:45] Maryam Pasha: Let’s get into it. Uh, Ben, thank you so much for joining us.
[00:02:49] Ben Williams: Thanks for coming, guys.
[00:02:50] Maryam Pasha: How are you feeling after your talk? That’s the first biggest question.
[00:02:53] Ben Williams: Uh, that’s a good question. I guess a lot of speakers would probably describe it as relieved. Um, [00:03:00] I’ve never practiced something so much in my life that I’ve been ready to just.
[00:03:04] Ben Williams: Never touch again. Yeah. So, but yeah, it, it was an amazing day. It was fantastic. I was really grateful to the team for the day as well. And now it’s come out, all the support from friends and strangers online has been, uh, really heartwarming. So feeling great about it.
[00:03:19] Maryam Pasha: Brilliant. For those of you, the very few out there who have not watched your talk, can you just give us one line, two lines on it,
[00:03:27] Ben Williams: I guess in a line or two.
[00:03:28] Ben Williams: My talk is all [00:03:30] about showing how. We can use AI to tackle challenges that matter for an environment by sharing the work I’ve been doing, which is working with coral reefs and listening to these habitats. So how we’ve been taking this tool and turning it to something you would never expect it could be helping us with.
[00:03:46] Ben Williams: Um, so that’s probably a talk in a nutshell. Brilliant. Yeah.
[00:03:50] Understanding coral reefs and climate change
[00:03:50] Ben Williams: Sorry to interject
[00:03:51] Ben Hurst: here, but can we take a step back? ’cause I don’t necessarily understand how coral reefs are linked to climate. Like obviously they’re [00:04:00] part of. Climate, right? Like they are in the, in the ocean, but like what role do they play?
[00:04:05] Ben Hurst: What impact do they have?
[00:04:06] Ben Williams: Yeah, totally. So. Coral reefs, like any habitat, they’re of course being impacted by climate change and they’re part of this giant biosphere that we live on. Um, but they are perhaps the most threatened ocean habitat that we have. So, bit of doom and gloom, but we’ll get it out the way at the start, I guess.
[00:04:25] Ben Williams: Right, right, right, right. So you’ve probably heard about things like coral bleaching where, uh, the ocean water [00:04:30] gets too warm that causes corals to bleach. They go this ghostly white. Mm. Um, a bit of a science behind that is that they have this. Symbiotic algae inside them. And when their temperature gets too hot, they’re expelled, but they provide corals with loads of air energy from photosynthesis, right?
[00:04:47] Ben Williams: So without them, after a few weeks, the corals will die. And we’re seeing this happen across the world, uh, at larger scales as climate change intensifies. So that’s where climate change and coral reefs are linked. And there’s loads of other [00:05:00] threats that are coming about because of climate change as well, like rising sea levels and increasing intensity and frequency of storms.
[00:05:06] Ben Williams: So it’s this pretty enormous existential threat for coral reefs at the moment.
[00:05:12] Ben Hurst: And is coral, uh, is it a, a plant or is it a rock? Like what’s the say? That might be a stupid question, but I don’t know. No, it’s not at, I just know not, not to touch them. That’s what I’ve been taught about. Yeah. Stay away. Yeah.
[00:05:22] Ben Hurst: As much as
[00:05:23] Ben Williams: possible. You’re, you’re right on both counts. Okay. So, well, I suppose I love it when I’m right on both count, not so. They’re kind of plant, animal and rock. So [00:05:30] they’re, the core thing is they’re like this coral animal, the polyp that is super simple. It’s basically got a, a stomach and some tentacles and it pulls food in and then they secrete the skeleton.
[00:05:41] Ben Williams: So that produces the reefs that we see. So these are normal structures that, you know, the whole great Barrier Reef is built by these organisms. Mm-hmm. And then inside them they have this algae. So that’s what gives them the color. That’s the plant. Yeah, that’s the plant component. So that also provides some energy by photosynthesis.
[00:05:59] Ben Williams: And [00:06:00] so they’re kind of this brilliant combination of the two. So you’ve got rock, plant, and animal all in the mix there. Wow. That’s pretty
[00:06:05] Maryam Pasha: incredible. Yeah.
[00:06:06] Ben Williams: I,
[00:06:06] Maryam Pasha: I love this idea that they like. Spell their, you said skeleton. The skeleton, yeah. That’s bizarre. It’s, it’s incredible.
[00:06:13] The importance of coral reefs
[00:06:13] Maryam Pasha: So in your talk, one of the ways that you talk about coral reefs that I love is you call them the rainforests of the ocean.
[00:06:19] Maryam Pasha: Mm. And obviously the rainforest is super important for us on land in terms of oxygen and in terms of being a carbon sink and stuff. I’m assuming that. [00:06:30] Coral reefs have this function in the sea around being like incredibly important for biodiversity and everything else.
[00:06:36] Ben Williams: Yeah, absolutely. So the kind of technical term we use is the ecosystem services that these habitats provide.
[00:06:43] Ben Williams: And so just some of the numbers are 25%, probably a bit more of all marine biodiversity can be found on coral reefs. No way. And when you look at the ex, seriously? Yeah. It’s, it’s enormous. And you, when you get on a reef, it’s, it’s [00:07:00] overstimulating the amount of animal life and plants and things you’re seeing here.
[00:07:03] Ben Williams: Uh, and that’s in this tiny area of the ocean as well. So it’s less than 0.1% of the ocean floor. It’s coral reef, probably much less. But it, it really punctures above its weight in the biodiversity. And so all that life, all this kind of. Ecological processes that are happening here. They provide all these services to millions of people, so it’s nearly half a billion with a B.
[00:07:25] Ben Williams: People around the world are directly reliant on coral reefs for [00:07:30] some kind of service, so that could be coastal protection from things like storms. So we see it in places where reefs have been lost or damaged. Those communities are gonna be far more impacted by storm damage when it rolls through. Or things like micronutrients and a protein source.
[00:07:47] Ben Williams: Uh, so particularly for childhood development in a lot of these countries, they, their primary micronutrients for many of those that they need come from the fisheries that are available on these reefs. Um, and then things like tourism in many [00:08:00] economies that’s, you know, done in the right way can be a really great way to generate income for them in a sustainable way.
[00:08:04] Ben Williams: Mm-hmm. So there’s all these kinds of. Services that we get from these habitats that really matter.
[00:08:09] Ben Hurst: You know why I find this so interesting here because I, based on your accent, my assumption is that you didn’t grow up. Near a coral reef That’s, I might be very involved. I think we’ve got a Sherlock Holmes over here.
[00:08:20] Ben Hurst: Yeah. Yeah. So, so how does, how does a person become like so interested and so invested in like what led you to Yes. Why do you care?
[00:08:29] Ben Williams: [00:08:30] Yeah. No, it’s a really good question and I. Uh, I don’t even know myself sometimes. Right. Um, but I guess it’s probably a cliche answer is I was, I was amazed by the natural world growing up.
[00:08:41] Ben Williams: So yeah, I grew up watching all these documentaries about the natural world and I, I also grew up, I. Kind of relating to that Greta Thunberg energy, you know, of we have this crisis that we’re not doing anything about. And so I wanted to be able to contribute in that space. That led to me, you know, [00:09:00] going the academic group studied biology and undergrad.
[00:09:02] Ben Williams: I had some amazing lecturers there. One of them, a long term mentor, Steve Simpson, who was working on co reef. He’s been on Blue Planet and lots of amazing things. Shout out Steve. Shout out Steve. Definitely, um, it, it just kind of. Came about through that route. Uh, they, they’re like this hyperverse ecosystem that to me is as close as we can get to an alien planet on the one where we live.
[00:09:23] Ben Williams: So it was this fascination and realizing how important, but how threatened these habitats are. That’s [00:09:30] how it came about for myself.
[00:09:31] Maryam Pasha: Incredible. Incredible. And I think one of the things that was so interesting as we hear more about your work and in the talk is, is that you layer all these things on top of this work, right?
[00:09:40] Maryam Pasha: So you’re not just a marine biologist looking at coral reefs, but you’re looking at the soundscape of a coral reef. And then we’re gonna also talk about ai. So before we get to the ai, we’re gonna build, right? So now we’re gonna talk about the idea. One step. Yeah. One step at a time. Soundscapes. So.
[00:09:56] Exploring coral reef soundscapes
[00:09:56] Maryam Pasha: We are gonna play for you some sounds from a coral reef, and then I want you to tell us what we are listening to.
[00:10:12] Maryam Pasha: And you played sounds like this in your talk as well, and I remember the first time I kind of heard these with you, I was just so surprised that there was so much life that you could hear. I mean, I just don’t associate sound with being under water.
[00:10:26] Ben Williams: No, I I think that’s exactly, you’ve hit on the point that the first.
[00:10:29] Ben Williams: [00:10:30] People say, often we get comments are like, what’s the point? Fish don’t have ears, but they do. You just, you can’t see them. Um, they look
[00:10:37] Ben Hurst: like yours. Yeah,
[00:10:38] Ben Williams: yeah, yeah. There’s so much going on. So this rainforest analogy, I think is a brilliant in this case as well, because, you know, in the rainforest you have all these birds, the mammals, the insects that you can hear, and it’s just exactly the same on a coral reef.
[00:10:52] Ben Williams: But it’s, it’s different. And of course. No one would ever have any reason to hear this normally. So most people, if they ever go on to a reef for the first time, maybe they’re snorkeling [00:11:00] or diving, it’s so kind of mind blowing. There’s so much going on, you don’t even notice. Yeah. You, you don’t notice that there’s all this sound.
[00:11:07] Ben Williams: Wow. But if you take a moment, you know, you maybe duck dive down and you. You hold onto a rock, not a coral or something, um, and you just stop and listen. The first thing you begin to hear is this kind of background crackle. That’s like, it’s like a campfire in the rain, and then you realize how loud it actually is, and then if you listen for a bit longer, there’s all these other sounds.
[00:11:28] Ben Williams: So those kind of [00:11:30] whooping sounds or grunting pulses, knocking. That is mainly coming from the fish that we have on these reefs. And I’ll be honest as well, most of these sounds, we don’t know what species is making them and why. So it’s totally a new area of science, but we’re starting to learn a lot more about it now that we’re putting microphones in the water and listening to these habitats.
[00:11:50] Maryam Pasha: So that’s what I was gonna ask you is how do you listen to these sounds as a scientist?
[00:11:54] Ben Williams: Yeah, so it’s pretty simple really. We’ve, we’ve got now low cost recorders. We worked [00:12:00] on developing some of these a few years ago called the Hydro Off. We can place these onto the reef and you can leave it recording for weeks at a time.
[00:12:06] Ben Williams: Wow. And it’s called passive acoustic monitoring. So it’s a whole field and you just leave it there listening and then. We, we pull that up, you download the data and that’s actually where the tricky part starts. We’ve got all this data so you can go through and listen to a, a small amount of it yourself, but this is where my work has come in trying to automate some of this process.
[00:12:27] Maryam Pasha: Right, right. The dead reefs, the bleached reefs that you talked about, [00:12:30] they’re pretty silent.
[00:12:31] Ben Williams: That’s it. So this is an emerging research topic, right? So we’re still learning all about this. Okay. But you know, if a reef has bleached in the last few weeks, we might still have a lot of the other organisms.
[00:12:41] Ben Williams: They’re there, they’re still making those sounds. Sure we could find that there is gonna be a change in the soundscape, but over time, if that reef doesn’t recover, if that reef dies, yeah, the soundscape is definitely gonna be impacted. Right? So when the coral is bleached and it’s still that ghostly white, that coral is alive, it’s just.
[00:12:57] Ben Williams: On its last legs. Right. Wow. So if the [00:13:00] water’s cool again and that algae can come back, the coral should recover and the reef will go back to normal. So
[00:13:06] Maryam Pasha: interesting. Oh wow. Wow. I didn’t know that. But as you said, you were leaving those devices down there for a really long time. And this is where we get to your talk, which I think is incredible.
[00:13:14] Maryam Pasha: Right. So tell us a little bit about how on Earth.
[00:13:18] AI in coral reef conservation
[00:13:18] Maryam Pasha: AI, because everyone is talking about ai. What are you doing with AI when it comes to this? This work?
[00:13:24] Ben Williams: Yeah. So AI is one of those things that is really revolutionizing this field. [00:13:30] And so it began where we collect all these recordings. We could collect weeks or months of audio on a single research expedition, and increasingly by sending ’em out to partners around the world.
[00:13:41] Ben Williams: So we have all this audio that no one could ever possibly listen to all of it. Mm. And so. We were looking for ways to automate the process of studying these sounds and learning what we can, and this began quite simply by kind of chucking the sounds of healthy reefs, just the, the raw recording, like that entire soundscape at an AI model [00:14:00] alongside those from degraded reefs.
[00:14:01] Ben Williams: And we found that. This model could actually do a pretty good job at identifying the difference between these. So if we gave it new recordings from a reef it hadn’t heard before in where we were working in Indonesia, it could tell whether this came from a he or degraded reef. Right? From just a minute of recording, we found we could apply this to challenges that mattered as well.
[00:14:22] Ben Williams: So. We work with Mars on that coral reef restoration program.
[00:14:26] Maryam Pasha: They are the, the, the backing supporting this [00:14:30] incredible restoration project that you’re working on out in Indonesia.
[00:14:33] Ben Williams: Yeah, that’s it. And, and around the world actually. Okay. And it is, yeah. Totally. When you mention this, people are confused. They don’t, they’ve heard of Mars, but they, they think it’s some other company or NGO.
[00:14:43] Ben Williams: Yeah. Um, but yeah, this, this is Mars who they own a whole load of brands. So pet foods are the biggest part of their business. Shiba being one of those. Mm-hmm. They have been doing a phenomenal job, uh, one of their kind of corporate responsibility projects in working on coral [00:15:00] reef restoration. I
[00:15:01] Maryam Pasha: do, I think one of the reasons I struggle with it.
[00:15:03] Maryam Pasha: In my head is, I’m like, but cats don’t like water. Yeah. I don’t know why. Yeah. Cats
[00:15:07] Ben Williams: like fish though. I think that’s starts spin. Okay. Okay. The cats gonna like, cool. Is the question. Yeah.
[00:15:13] Maryam Pasha: So anyway, this is a complete tangent, but I realize that that’s why my brain trips up on it because I’m like, but the, but the cats don’t like the, the water, but they do like the fish.
[00:15:21] Maryam Pasha: Going back to this incredible technology one, one question. Normally before you bring in something like ai, do people [00:15:30] just li is it just like manually, like people listen to the recordings and then figure out what’s going on?
[00:15:35] Ben Williams: Yeah, that’s it. So I, I can link those two together. The first set of recordings I was involved in gathering were from, I.
[00:15:41] Ben Williams: Hope Reef or before Hope was built actually at this restoration project in Indonesia, which Mars is supporting. And Tim, another mentor of mine, he spent many hours listening to the recordings and it takes ages, right? If you want to listen to an hour of audio. It’s probably gonna take you all day to listen to it and count the different [00:16:00] sounds, figure out what these are because they’re all new sounds.
[00:16:01] Ben Williams: We don’t, we don’t hear it. You know, like we would in birds bioacoustics and go, oh, that’s a gold finch. Or, you know, that’s a robin, right? We, we don’t know what these are. So you have to find the sound for the first time. Figure out what the different ones are, and then how often are we hearing these? So that was initially how it started.
[00:16:18] Ben Williams: Kind of manual process. Very intensive. Yeah.
[00:16:20] Ben Hurst: Yeah, that’s tough. And you mentioned also building Hope Reef, which is where your work is based now. Mm-hmm. And how do you build [00:16:30] a a, a
[00:16:30] Ben Williams: reef? So I could never do it justice in one answer, so I would definitely, if you want to know more, look it up online. So building coral.com mm-hmm.
[00:16:39] Ben Williams: Is the website where it can tell you all about the process. But in a nutshell, you have this. Degraded reef, it’s destroyed.
[00:16:46] Coral reef restoration techniques
[00:16:46] Ben Williams: It’s all kind of rubble, right? That doesn’t recover. It becomes ecologically trapped, as we call it. If we don’t intervene, because the rubble’s just loose. If a new coral settles on it, it turns it, it turns in the waves and the tides, it doesn’t have a way to grow back, right?
[00:17:00] Ben Williams: So what you can do is you can put out, they call them these reef stars with loads of corals attached to them. Link them up, and they kind of secure the, the bottom habitat. And then the coral begins to grow, so it’s given it this secure substrate to grow from and they can carpet huge areas, you know, several hundred square meters in a week, and after just a few years, it’s stunning seeing how it’s come back.
[00:17:23] Ben Williams: You can swim over this and you kind of peek through. You can maybe see like these reef stars at the bottom peaking out, but they’re really hard to [00:17:30] see in a few years later is. It’s gone.
[00:17:32] Maryam Pasha: Okay, so we are building these weaves. We’re looking at this, but the data is overwhelming.
[00:17:36] Introduction to AI in coral reef restoration
[00:17:36] Maryam Pasha: So now we get, we finally get to the ai.
[00:17:40] Maryam Pasha: This feels, talk us through this rationale because I think that then people are not immediately saying, yes, I love restoration. I’m into it. I’m listening to a coral reef. Let me use ai.
[00:17:51] Ben Hurst: Let me plug chat. Gpt in, like, I feel like people just know stuff about the capabilities of ai. Maybe I just need to do some research on ai, but tell us [00:18:00] like how do what, how do you arrive at that conclusion?
[00:18:02] Ben Williams: Yeah, and I guess that’s the thing.
[00:18:03] AI applications and challenges
[00:18:03] Ben Williams: It, it’s one of those messages I want to share is you, you think of chat GBT or you think of image generation or something when AI is mentioned to you, but it’s this tool we can use for so many different problems. So this project came about for myself, it was kind of.
[00:18:19] Ben Williams: Serendipitous in that I was set to be working on a project. Covid happened that completely fell through. And so I was looking at this other data set we had where my colleague had worked on it. [00:18:30] He’d listened to all of these recordings in this really labor intensive process, and I’d actually worked on a similar problem in a previous coursework when I was an undergrad.
[00:18:40] Ben Williams: Where we, instead of coral reefs, we had human patients who had been diagnosed with heart disease or as healthy and all these different measures from them. So like blood pressure, cholesterol levels, those kind of things. And we trained a very simple AI model to be able to diagnose new patients. And so I realized actually [00:19:00] these problems aren’t too dissimilar, but instead of patients we have.
[00:19:03] Ben Williams: Coral reefs and instead of those kind of health measures, we’ve got acoustic recordings. So that’s where I set about on this problem. And it, it worked really well the first time we gave it a go. And so we’ve been running with it from there and trying to build this field and try all the different kind of nooks and crannies we can explore here to, to make use of it.
[00:19:21] Ben Hurst: That’s like a, that’s such, I mean, such a good way of using ai. ’cause there’s also a big, a big conversation at the moment about ai. Maybe being a bad thing or like [00:19:30] using too much energy or something. Yeah. But I feel like if you’re using it for this, then that makes sense, right? Like that maybe this is a good way Yeah.
[00:19:36] Ben Hurst: Rather than Absolutely. Generating studio oji images, you know? Yeah, yeah, yeah. I dunno.
[00:19:41] Ben Williams: Yeah. I think that’s the thing. It’s, it’s one of those technologies that’s gonna be completely pervasive. It’s gonna change the world we live in, right? And it’s trying to make sure some of that gets captured for good.
[00:19:51] Ben Williams: Not just for, like you say, making fun images of you or your friends doing whatever. So that’s what I’m hoping we’ll see more and more of.
[00:19:59] Ben Hurst: Yeah. [00:20:00] And so from this, from this, like your undergrad research and then repurposing that learning into this research and this work that you’re doing now, was that, you said that that was like a pretty simple.
[00:20:11] Ben Hurst: Process and you just train the AI and it and it works. And it’s worked. It worked the first time and it’s carried on working. Have there been any surprises? Was there anything that you like came across that was like, oh my gosh, I didn’t realize that that was gonna happen, or We’ve discovered something completely new?
[00:20:25] Ben Williams: Yeah. Well, I guess the process in itself was certainly completely new, but it’s not without its challenges. So we [00:20:30] do find this model that we trained there. It works at this, this one time, this one place. If we took it somewhere new, like the Great Barrier Reef from mm-hmm. Last year it’s gonna fall head over heels.
[00:20:40] Ben Williams: So these models don’t transfer to new data sets. Okay. So we’ve been trying to work on approaches that can actually help us solve that problem. And so a big surprise in some research, this was where the time I spent with Google DeepMind came in, was we were trying to capture some of the learnings from.
[00:20:59] Ben Williams: Other bio [00:21:00] acoustic fields where they’re far better studied. Mm-hmm. We found that if we actually trained a model on audio from birds, which we have millions of recordings of people have been listening and documenting birds for like long decades with Right. People love that bird, love bird, that’s it. And so we realized we could actually capture that.
[00:21:18] Ben Williams: And train our model on a huge abundance of bird data with a sprinkling in or a, you know, a healthy dose of coral reef audio and optimize it for the coral reef challenge, but really building on the [00:21:30] foundations that other bio acousticians have managed to achieve in this bigger field. Mm-hmm. So that was a real surprise and it worked really nicely.
[00:21:37] Maryam Pasha: And, and what, in terms of the restoration work, what this allows you to, to understand right, is, is. If it’s working.
[00:21:47] Ben Williams: Yeah, that’s, that’s it. So we want to understand if restoration is working, are we bringing back what we would call a fully functioning ecosystem? So something that’s comparable to a healthy reef [00:22:00] or something that’s not quite there yet, or is different that’s reemerging in a different way.
[00:22:04] Ben Williams: And so we’re still developing these fields, but we’re finding we can. Here indicators of many of the processes on these reefs in the soundscape that we want to see if they’re recovering. And so we’ve been doing this in Indonesia, but also with a handful of partners around the tropics restoring their coral reefs so they can see is their approach making a difference?
[00:22:24] Ben Williams: Right? Can they evidence that that’s really important to a lot of these projects? You know, they’re trying to raise funds or. Maybe they’re [00:22:30] a tourism operating. We wanna show their customers that they are restoring the reefs. Um, so that’s how we’re being able to use this tool to support coral reef restoration.
[00:22:39] Community involvement in coral reef projects
[00:22:39] Maryam Pasha: So Ben, one of the things that we talked about a lot when we were working on your talk, just ’cause I’m always interested in this and we, I was kind of, we were discussing it, is the role of local communities in your work, because obviously we’re sat here in East London, right? The coral reef is not here.
[00:22:54] Maryam Pasha: Allegedly. Yeah, allegedly. And as you said, there are loads of local communities all around the world who [00:23:00] depend on these reefs, but also have been stewards of these ecosystems for, you know, centuries if not thousands of years. Mm-hmm. I’m curious to know, as a scientist from the uk, how do you work with both scientists and other local experts and people who, who live and whose livelihoods are on the reef in these kinds of projects?
[00:23:21] Ben Williams: Yeah. It’s a really important question and it’s, it’s something I think science has improved on a lot over the last. Decade, I’m sure there’s still a long way to go, is this [00:23:30] concept of parachute science where, you know, western scientists comes in, collects some data, flies back home, publishes paper, and there’s no engagement there.
[00:23:38] Ben Williams: Right. Um, but that’s where with projects like this one, through listening to the reef, it’s a really powerful way to. Engage many different layers and groups of people. So we’re able to send recorders to people around the world who have this problem they want to solve. Mm-hmm. So in our case, a, a big bottleneck was being able to monitor reef restoration that lots of [00:24:00] different projects have been implementing, uh, because it’s really costly and time consuming to restore a reef, let alone go, and then do the impact reporting and the monitoring.
[00:24:10] Ben Williams: So. Us as you know, scientists sat here at University College London or where, anywhere else, you know, we might be working on technologies that can help with some of these problems that these local groups are trying to solve. So by collaborating together, we can produce this new technique and test it out on their reefs, working with them, and then give them the [00:24:30] results so they can report on that impact.
[00:24:32] Ben Williams: And it’s, we’ve been doing that for restoration, but there’s definitely other use cases where I think we want to start. Tr finding new partners who can benefit from this as well. And we’re always open to people who have ideas and projects that could use this technology as well.
[00:24:46] Maryam Pasha: Right.
[00:24:46] The issue of bomb fishing
[00:24:46] Maryam Pasha: So I think this brings us nicely to something you kind of mentioned at the end of your talk.
[00:24:51] Maryam Pasha: We didn’t have enough time to like really get into it, which is this problem of bomb fishing, which is something I hadn’t heard of before. Mm-hmm. Um, will you tell us a [00:25:00] bit about it?
[00:25:01] Ben Williams: Yeah. So I. I guess firstly, bomb fishing. Basically it is what it says on the tin. So fishermen go out on the water and with these explosives mostly made often of homemade components, throw them onto shoulder fish and it’s a devastating and fast way to kill all these fish.
[00:25:19] Ben Williams: Scoop ’em up is a very efficient way to catch and kill a lot of fish. I do that in um.
[00:25:24] Ben Hurst: Breath of the wild, the Zelda game. Sorry. For context. Like it’s a, it is a really efficient way of [00:25:30] fishing. Well, I’m glad you can relate first time. Yeah, yeah. Well I wouldn’t do it in real life. Important to mention it.
[00:25:35] Ben Hurst: Yeah.
[00:25:36] Ben Williams: Yeah.
[00:25:36] Ben Hurst: I
[00:25:36] Ben Williams: haven’t heard that application before, but yeah. Like
[00:25:39] Maryam Pasha: what on earth you doing? I was, we need to chat. Yeah. You do Your weekends,
[00:25:46] Ben Williams: the weekend climate confession. No, no, no. Yeah, this week’s climate confession. Um, but yeah, that’s it. And so it’s. A very quick and fast way to kill a lot of fish, but it also kills the habitat where they live.
[00:25:59] Ben Williams: So [00:26:00] the coral reefs and this bombing will happen again and again over years and decades and it will decimate that reef and reduce it to those rubble fields I mentioned earlier that failed to recover. Um, and it’s a big problem in, we did a. A review recently and found 20 countries in the world across the last 20 years.
[00:26:16] Ben Williams: It’s been happening in just from published literature. That’s probably more out there. And in many of these, perhaps most, it’s found that often there’s a bit of a cost barrier. So often it’s the more middle income or wealthier members of those communities can [00:26:30] access the illegal supply chain to get the parts needed to make these.
[00:26:34] Ben Williams: And often it’s the more vulnerable community members who. Don’t have the resources to this who go out and use traditional fishing methods who are then losing their livelihood because the reef is gone, the fish are gone. So it’s, it is impacting these people in lots of different ways.
[00:26:47] Maryam Pasha: Yeah. So it’s like the most vulnerable parts of society, and once again, suffering.
[00:26:51] Maryam Pasha: Yeah, it’s one of those
[00:26:53] Ben Hurst: stories. Yeah. Yeah. And are people, people, I mean, you might not know the answer to this, but the people who are doing this, doing it because they. Don’t [00:27:00] understand the impact of it, or do they know and they just don’t care that they’re destroying the reef.
[00:27:04] Ben Williams: It’s really nuanced and it changes around the world and you probably have to talk to everyone, but what we often hear is that people are aware of the damage it does, but often the sea is considered so large that you know, the small impact you are having isn’t gonna add up.
[00:27:19] Ben Williams: I mean, we all live like that, really don’t we? Many aspects of our life. It’s like the last thing reefs need in the face of climate change is like people throwing bombs on them. Yeah. Yeah.
[00:27:29] Maryam Pasha: Do [00:27:30] you, but this p but this practice is illegal, right? Like it’s not that this is something totally fine, like what people are doing when they’re fishing in this way is illegal.
[00:27:38] Ben Williams: Yeah. That’s it. It’s, it’s illegal in every country and it’s, it, it happens in loads of places, but a blind eyes turn to it or often there’s just no way to document where it happens and how often. It’s just rumors and things being passed around communities and groups. So we can’t know where it’s a big issue.
[00:27:56] Ben Williams: If some kind of focus or attention is needed, and often that makes it easy to [00:28:00] turn a blind eye right. When there are other problems in an area as well.
[00:28:03] Maryam Pasha: Right? Because it’s like how vast is the ocean And the ocean,
[00:28:07] Ben Williams: yeah.
[00:28:07] Maryam Pasha: How are you gonna detect it? So I, I’m curious to know then, how does all of this play into your work?
[00:28:13] Ben Williams: Yeah, so this was another one of those where. It kind of, it came about once we developed this first approach using ai, it seems so obvious that this had to be the next thing we wanted to take on, because I’ve heard it firsthand in the water. [00:28:30] So you’ll be on your dive or your snorkel. And you’ll hear the bomb go off and it can be so loud, it shakes you out, your skin, but you put your head up on the surface and you can’t even see the boat that’s responsible.
[00:28:43] Ben Williams: And that’s because sounds travel so far underwater. Oh yeah. Yeah. So it travels. 4.3 times faster underwater, meaning it goes great distances. So we can hear these bombs probably as far as 50 kilometers away. Wow. So similar technology has been used for poaching, for [00:29:00] gunshot detection, where in a rainforest, you might be lucky if you can hear a gunshot go off a kilometer away.
[00:29:05] Ben Williams: Mm-hmm. But we can hear this over a huge distance. So it made us realize, well, why don’t we start looking at the. Data we have in these audio and start putting down recorder deliberately to start seeing how much bomb fishing that was happening. Mm. So in one area in the Indo-Pacific, we worked with some partners.
[00:29:20] Ben Williams: We can’t reveal who they are at this stage, but hopefully in the future they’ll get the credit they deserve. Mm-hmm. And we found from a single recorder. Using the AI modeler helped build for the team [00:29:30] they detected and then verified listening to the results over 4,600 incident of bomb fishing from
[00:29:35] Ben Hurst: 4,600.
[00:29:38] Ben Williams: That’s it. And that’s from one recorder in six months. Oh my God. So it’s, it’s scary numbers.
[00:29:44] Ben Hurst: That’s a lot. So
[00:29:44] Ben Williams: you,
[00:29:44] Ben Hurst: you, you’ve become. The, the bomb fishing police accidentally detectives. You did? Yeah. Yeah. Detectives, yeah. Not police. We’re still,
[00:29:52] Ben Williams: but detectives at this stage, and so that’s crazy. I, I really hope this can be a tool that either us or others can start developing and [00:30:00] using elsewhere, right.
[00:30:01] Ben Williams: To solve that first problem that I mentioned where we can find out, well, where is this happening? You know, in a nation’s waters, where are the hotspots? Where are the problems? Yeah. How much does it happen? And then if interventions are put in place, so that will need a combination of enforcement. But also the social aspects.
[00:30:18] Ben Williams: So alternative livelihoods and awareness campaigns that will allow people to move away from this practice sustainably. Can we then measure the reductions that hopefully those interventions can have, [00:30:30] and one day, I hope, report some positive numbers or. Negative numbers, I guess, in terms of happening, but hopefully we can use this combination of technology and that social intervention hand in hand to solve the issue.
[00:30:47] Maryam Pasha: I think that’s really cool. And I, and I, I guess I really love the idea that you kind of have gone on this journey mm-hmm. From, you know, in, in all of these different ways.
[00:30:55] Future directions and innovations
[00:30:55] Maryam Pasha: And I, and I, and I’m just curious to know, like, so bomb fishing is obviously one of the things that you are looking at now in your, in your lab and with your.
[00:31:02] Maryam Pasha: You know, collaborators. Are there other kind of places you’re hoping to go in the future with this work?
[00:31:08] Ben Williams: Yeah. Well, how long have you got? There’s, there’s, uh, me personally, a number of directions, especially this bomb fishing angle. We’re also looking at how we can produce these giant. Digital twins, we call them of habitats.
[00:31:21] Ben Williams: Yes. Tell us
[00:31:22] Maryam Pasha: about these.
[00:31:22] Ben Williams: So digital twins, these, these are stunning. I’ve got a friend we’re working on this called Sergi. If you look up wild flow, you’ll find his website and he’s [00:31:30] got these stunning models of this enormous reef area where you go on the reef, you collect a ton of pictures and you can produce this giant 3D model.
[00:31:37] Ben Williams: And then one of the master’s students in Indonesia we worked with just published her lead off paper. So shout out to Linda showing that we can use these models. To track restoration progress, so on that same system. Um, so that’s another angle, but you’re
[00:31:50] Maryam Pasha: gonna have to explain though to, to, to Ben. What is a digital twin?
[00:31:55] Maryam Pasha: Oh,
[00:31:55] Ben Hurst: do you already know what it is? Yeah, it’s just me. It’s just me and you guys listening. [00:32:00]
[00:32:00] Ben Williams: What is a digital twin? Yeah. It’s so easy to reel off all the jargon, isn’t it? But it’s like this giant 3D model. So I guess you could have it in, you know, in Zelda or whatever game. Though you can look around on, you can explore.
[00:32:14] Ben Williams: It’s a photorealistic 3D twin of the reef and the often the digital twin, if you want to be really technical, it will have some live component to it as well. Maybe data coming in from the field. I hope we can get there maybe with acoustics soon as well. So that’s [00:32:30] one area that another one I’m really interested in looking at.
[00:32:33] Ben Williams: And then there’s colleagues doing loads of amazing stuff. So here at UCL there’s a huge project called the Bio Health Project, where research has been working with groups across the Mass MA in Kenya, in Borneo, in Nepal, to look at the impact of national parks and conservation zones and. The communities, again, they’re living with those, so those that live within each of these and using camera traps and AI where, so these camera traps and animal walks in [00:33:00] front of ’em and takes a picture nice and easy.
[00:33:01] Ben Williams: We collect millions of images, so AI can help us study those and finding amazing findings like what’s the impact of cattle grazing verse sheep and goat grazing in Kenya, for example. Yeah. All kind of research questions that. Often tie in back into climate change, into human health. That’s where a lot of my colleagues are working as well.
[00:33:21] Ben Williams: So it is like just this nexus of things that can be studied. I do have one
[00:33:24] Maryam Pasha: question because we’re in your, in at the UCL East, so you tell us about the bats very quickly.
[00:33:29] Ben Williams: There’s, there’s a [00:33:30] lot of bat stuff. So my, my supervisor, my lead supervisor for my PhD, Kate, kind of cut her teeth on bat. So I think she is,
[00:33:37] Maryam Pasha: we won’t have to have, we won’t have to have.
[00:33:39] Maryam Pasha: That woman on,
[00:33:39] Ben Williams: she’s definitely worth it. And Kate seems to know everything about everything when it comes to the environment and human health, so would recommend. But um, yeah, we’ve got recorders out on the Olympic Park where our office is based listening out for bat activity. Um, and it’s a great way to kind of test new technology that’s being developed and they’re looking at the impact, you know, the Westham [00:34:00] Stadium here.
[00:34:00] Ben Williams: Yeah. How does that change bat activity when there’s a game going on? AI is helping find the bats in those recordings and see what’s happening. So yeah, lots of crazy stuff.
[00:34:11] Ben Hurst: Love that. So interesting. I love, this is, yeah. Um, I re-watched Black Panther. The other day and it’s like, um, you know that bit where they, they have the model of the cars when they’re trying to do the mission anyway, but it sounds like that, it sounds like Wakanda in the Sea, which is Oh yeah, I know exactly the bear.
[00:34:26] Ben Hurst: That’s where we wanna get to. We’re
[00:34:29] Maryam Pasha: [00:34:30] brilliant.
[00:34:30] Conclusion
[00:34:30] Maryam Pasha: Is there anything else that you think we should be listing or people should be looking into if they’re interested in any of this work?
[00:34:38] Ben Williams: Yeah, I think it’s, you know, it’s a difficult question. People want to do so much all the time, which is amazing. And we can all only.
[00:34:45] Ben Williams: Do so much individually, and I think often it’s easy to pin environmental or climate changes. Onto the individual, which is not right. We can all play our part, we can all make a difference. So people are probably well aware of the changes they can make. But directly, if you want [00:35:00] to learn more about coral reefs, there is, there’s my talk online.
[00:35:02] Ben Williams: Please check it out. Yes, definitely.
[00:35:04] Maryam Pasha: Number one,
[00:35:04] Ben Williams: there’s, um, building coral.com is website. There’s a project called Calling in Our Corals, which is another collab with Google where you can learn all about coral reef sound and then contribute some data yourself. So it’s a citizen science project Wow. Where you listen to recordings and help us find.
[00:35:20] Ben Williams: New sounds within those that is,
[00:35:22] Maryam Pasha: we love citizen science. Yeah, yeah, yeah,
[00:35:24] Ben Williams: yeah. So that’s a big one. And then also, I always like to say younger listeners or someone looking for a change as well. I [00:35:30] think well, well, I’m a big believer that one of the differences you can make is choosing what you do of your career and where you’re going.
[00:35:35] Ben Williams: So maybe that’s research like me, or you want to get into the policy aspects or sorting out the supply chain side in some business or industry you work with. So you can make such a difference by all that years of life you dedicate working if you can. Turn that onto something useful. I think that’s where big change can be made as well.
[00:35:54] Ben Williams: Yeah. Yeah. Great advice. You can be the next Batman
[00:35:58] Ben Hurst: is not what you were saying [00:36:00] actually. Yeah. Maybe cut.
[00:36:03] Maryam Pasha: Thank you so much for joining us mm-hmm. And for letting us kind of get in and pick and ask all the questions we wanted to ask about your work. Um, we could have, we could have gone, I mean, we could do a, maybe we’ll do like a whole episode just on the AI side of things.
[00:36:15] Maryam Pasha: Mm-hmm. ’cause I feel like there’s so much cool stuff around how you made these models, so maybe we’ll come back and do that. But thank you mm-hmm. For, for telling us about your work and for hosting us here.
[00:36:25] Ben Williams: Yeah. Thank you both. It’s been amazing to have you and get this chance to talk about it more. Um, I’m always [00:36:30] here.
[00:36:30] Ben Williams: So yeah, reach out.
[00:36:31] Maryam Pasha: We’ll come back
[00:36:31] Ben Hurst: to Visit Facts. I’m gonna have some recordings for you soon, so be prepared. Can’t wait. Remember, stay curious.
[00:36:40] Maryam Pasha: Thank you for joining us this week. We really hope you’ve enjoyed this episode.
[00:36:43] Ben Hurst: If you did, please hit the follow button to make sure that you get next week’s releases.
[00:36:48] Maryam Pasha: And why not reach out with your very own climate confessions?
[00:36:51] Ben Hurst: While you’re at it, please leave a rating and a review.
[00:36:55] Maryam Pasha: Now, let’s shout out to our fabulous team behind the pod
[00:36:58] Ben Hurst: executive, produced by [00:37:00] Josie Colter,
[00:37:00] Maryam Pasha: produced by Ben Beheshty.
[00:37:02] Ben Hurst: Music, also by Ben Beheshty,
[00:37:04] Maryam Pasha: communications by Tara Cooper and Izzy Gladston.
[00:37:07] Ben Hurst: Curation by Deesha Chandra,
[00:37:09] Maryam Pasha: hosted by Ben Hurst.
[00:37:11] Ben Hurst: I’m Maryam Pasha, and remember, stay curious.
