Ai & the Environment (Projects Episode)

Two projects loosely under the umbrella of the environment.

This episode was recorded on May 10th 2023

Calum Corrie Imrie - Preserving Marine Life in a Shipping World: AI to the Rescue (PREVAIL)
 
Karl Dickinson - AgriTrust

Podcast production by boardie.com
 
Producers: Louise Male  and Stacha Hicks
 
 

Podcast Host: Sean Riley

The UKRI Trustworthy Autonomous Systems (TAS) Hub Website

Transcript

Living With AI Podcast: Challenges of Living with Artificial Intelligence

This podcast digs into key issues that arise when building, operating, and using machines and apps that are powered by artificial intelligence. We look at industry, homes and cities. AI is increasingly being used to help optimise our lives, making software and machines faster, more precise, and generally easier to use. However, they also raise concerns when they fail, misuse our data, or are too complex for the users to understand their implications. Set up by the UKRI Trustworthy Autonomous Systems Hub this podcast brings in experts in the field from Industry & Academia to discuss Robots in Space, Driverless Cars, Autonomous Ships, Drones, Covid-19 Track & Trace and much more.

Season: 3, Episode: 7

AI & the Environment (Projects Episode) 

We look at two TAS  projects loosely under the umbrella of the environment.
  
 Preserving Marine Life in a Shipping World: AI to the Rescue (PREVAIL )- Calum Corrie Imrie  
 
 AgriTrust - Karl Dickinson

This episode was recorded on May 10th 2023


Podcast production by boardie.com
 
Podcast Host: Sean Riley

Producers: Louise Male  and Stacha Hicks

If you want to get in touch with us here at the Living with AI Podcast, you can visit the TAS Hub website at www.tas.ac.uk where you can also find out more about the Trustworthy Autonomous Systems Hub Living With AI Podcast.

Episode Transcript:

Sean:                  Welcome to Living with AI, the podcast where we meet to look at how artificial intelligence is changing lives and altering society and looking at things like how our personal freedom is affected and the impact it has on our wellbeing. 

                            Today we are looking at autonomous systems with respect to the environment which obviously affects everybody. This is season three of our podcast and there are plenty of back episodes for you to binge on. Links will be in the show notes, if you search for TAS Hub I am sure you will be able to find it.

                            We are recording this on the 10th of May, things happen very quickly and change very quickly in the world of AI as am I sure everybody that’s listening to this is aware, so it’s always very useful to work out when things are recorded.

                            So today’s episode is a project episode where we are looking at a couple of projects that the TAS Hub has funded and are grouped around that theme of the environment. So we have got Calum and Karl. What I will ask the contributor to do is just give us a brief introduction if you like, the name, rank and serial number. And then hopefully after we have met everybody we will be able to hear a little bit more about the projects and then afterwards we will be able to have a discussion about them.

                            So just for no other reason than you are on the left of my screen, Karl could you kick us of please, name, rank and serial number?

Karl:                   Yeah of course. Hi everyone, my name is Karl Dickinson and I am a research assistant from Newcastle University working on the AgriTrust project. We are a team of seven, led by Professor Nagaraja and we are working on an agricultural test bed to examine trust between IoT technology and its use in agriculture.

Sean:                  Fantastic thanks very much. And Calum?

Calum:               Hi thank you for having me. So my name is Calum Imrie and I am a research associate at the University of York and I am based at the Sharing Autonomy International Programme. So my main areas of research is looking at the safe deployment of robotics and autonomous systems. And one of the strands has gone into this TAS Hub project called Preserving Marine Life in a Shipping World, AI to the Rescue or PREVAIL for short.

Sean:                  Fantastic. And yeah I mean both obviously extremely important and worthy things to be looking at with AI. So let’s dive in a bit deeper with you Karl about the AgriTrust project.

Karl:                   Yeah so with AgriTrust as I said it’s a project examining the use of IoT technology in agriculture. So first of all what is IoT technology for those that don’t know what it is. So IoT technology is physical devices like vehicles or appliances or any object that concerns sensors and software and the ability to connect to the internet to share data and communicate with other devices.

                            So as an example, in agriculture sometimes you use automated sprinkler systems to water the crops basically. So you will have a small control system like taking a reading from a soil moisture sensor and use it to activate these sprinklers to give the crops the right level of moisture. So just in this basic example you can see how farmers are putting their trust in this simple automated system to water the crops.

                            But for example where that moisture sensor is not giving the right value for example, if the sensor is saying the soil is too dry but it’s not, the farmer or the automated system will over water the crops and damage the crops. It has got like serious implications for putting their trust in IoT technology.

                            So basically at a higher level I have been working on developing a sensor box which we have called the squirrel box and use this as a tool to investigate trust in IoT and agriculture. So the goal was to create devices that we can put outside and on farmers’ fields and take measurements of different soil and ambient perimeters, such as soil moisture, the temperature, the pH, ambient light, ambient humidity and things like that. And being able to transmit this data across to the farm to the farmer and use this tool to ask for trust, so trust not only in the data but trust, farmers putting trust in this technology. 

                            So we have got seven sensors, two soil moisture sensors, a temperature, pH, NPK, an ambient light sensor, which I said, but we needed something to read this data and transmit it across the farms basically. As you know farmer’s fields they can go on for several miles, they are not just in a small contained area so we couldn’t use Wi-Fi for that because that runs based on metres not miles. So the controller that we were using users LoRa which is Long Range Radio so we can transmit the data using long range radio and the farmer can read that and see all the useful information on the farm. 

                            But by creating these squirrel boxes, what we’ve called them, we can then split that down to three work packs if you like. So the first part which I was mostly involved in, building these sensor boxes, testing them and getting them ready for deployment. But Professor Richard Hyde and Doctor Peter Craigon who are part of this project, have been focusing on the stakeholders who are putting their trust in this technology such as farmers. And looking at what is holding them back from adopting smart IoT technology to control their farms.

                            So they have been looking at studies and I don’t know if you have come across the Moral IT cards, the ethics and research cards? So Craig and Richard have developed these cards which are a tool that explores issues around the ethics and focusing on social and economic and technical challenges. So they have been focusing on those implications. 

                            There is also, so we have got Dr Mujeeb Ahmed and he has been focusing on authentication methods for the data. So for example how can you trust that a sensor measurement is correct? So let’s say you had a pH sensor saying that your soil is very alkaline and it comes up with a pH reading of nine. So then the farmer might go out and say all right I had better do something about that and put some fertiliser out and make it a bit more acidic. But if that data is incorrect or it’s wrong, then the farmer can potentially destroy all of his crops and cost him a lot of money and serious impact to his business. 

                            So Dr Mujeeb Ahmed has been looking at how can we verify the correctness of a sensor. So when we deploy these boxes we wouldn’t just have one sensor box in an area we would maybe have five and several boxes. So it’s looking at, so for example if one box is saying nine as a pH value and the rest is saying five and a half, then you would be like right it must be five and a half and not nine. So he has been looking at techniques such as that and more complicated techniques to verify the correctness of the sensor data.

                            And the other part of the project which Professor Shishir Nagaraja is looking at is preserving the confidentiality of this data. So imagine a farmer adopts this technology and is getting fantastic results so he is optimising his resources. He has got optimal crop growth that has improved the yield that could be very valuable information for a competitor. So if the data is insecure or unsecure sorry, and you have someone nearby the farm who can pick up and read this unprotected or unencrypted data, they could use that to, well benefit themselves if they have got their own farm, or even sell this data on and make money. 

                            So that’s the other side of the project we have been looking at as well. I have probably gone on for long enough.

Sean:                  That’s great, no thank you. It gives a real sense of what you have been doing with the AgriTrust project. Can we listen to Calum talking about PREVAIL then please?

Calum:               So with PREVAIL, maybe I should start off and say that this is primarily a feasibility study. And hopefully it will become clear why we are starting off with this feasibility study rather than pushing straight on to the deployment of hopefully solutions that we have been thinking about in our group.

                            So to start off with the problem, ships are great there is no denying that. They are able to transport almost any sort of goods and are environmentally low impact compared to air travel for example. However, this huge increase of ships has put a massive strain on the ocean to almost basically catastrophic levels.

                            And what’s kind of terrifying is there is still no hard numbers on the full extent of these ship strikes with whales. Primarily because when these massive ships collide with a whale, a) they are not maybe aware that they have actually collided with said whale. And then b) if the whale does get shrugged off the front of the ship it will sink and nobody will be able to see to report it.

[00:10:04]

                            In fact one of the only good ways of actually getting a good idea of these are stranding’s so when whales are like, you know, pushed up onto the beach via the waves. And so for example a hotspot for these strikes are in the Hellenic Trench in Greece where they find at least fifty per cent of the whales have had some sort of collision with a ship based off the injuries that they have observed on the whales. So that’s kind of one aspect of the problem there.

                            And what’s kind of interesting was when we have been looking into it more and more in terms of okay what can we actually do, there is very few sort of actual effective mitigation techniques. So the first one is rebooting, so you know if you spot a whale then they can use what is known as a dynamic management area which is we have spotted a whale and now we have created this radius so please don’t go in here. We think there is going to be a whale or at least more than one whale present. 

                            And the other technique is slowing down.. So they believe that about twenty-two knots or something like that will be about ninety-nine per cent of just outright killing the whale. And then they want to reduce that to about ten knots or so to really reduce that probability. 

                            So that said those are really the only two effective techniques out there. And what is kind of interesting is, sorry not interesting, what’s horrifying sorry, horrifying that’s the word I want to use, is that both of these are voluntary techniques, right. There is very few parts of the world that will actually enforce this.

                            So we are kind of trying to think of a way on how can we I guess bring up those sorts of voluntary speed reductions right or those voluntary sort of rerouting actions. And there is a really good example of that that maybe kind of highlights this a bit more is there is actually an AI solution out there that is operating and it’s got quite a bit of nice media press about it, it’s called Whale Safe, along the Santa Barbara coast. 

                            And they deploy a few techniques. They use hydrophones as their primary sort of selling point here where they use these fancy convulsion neuro networks to essentially look at these underwater microphones and the spectrograms that they are able to get and then say is there a whale, is there not. And if you get a few of these hydrophones or you get them all in like one sort of vertical column, you are actually also able to pinpoint the position of the whale as well.

                            They have also got people on board other ships or on the coast with these apps able to report it. And you have also got these sort of GPS based stuff like the tagged whales, you know, via these marine biologists on the shore.

                            So they have been operating for about a year and a half now I think they have been in operation for and they claim that there has been no whale deaths at all. And to be clear I don’t think they are lying or anything like that, I want that to be clear. I don’t think that’s, it’s just this is what they have been saying and that’s fantastic news, you know. But when I read a bit more into it the marine biologists in one unison kind of just shrugged and they said that’s fine. The problem is this is not scalable to the entire ocean, you know it’s only this Santa Barbara coast.

                            And also from the interviews that they’ve done with captains, this is a bit more anecdotal evidence, there is that lack of trust. So if the captains get pinged on their phones that ah there is a whale nearby, unless they can physically see it they are not going to slow down. 

                            So that is kind of another interesting avenue that we have been really talking about within our group is deploying AI techniques, there are quite substantially good techniques out there to use. The question is how do we actually bring it forward and actually say yes this is trustworthy and this is the evidence that has been able to get to this. But also for the maritime domain that might not of course be AI or ML experts.

                            So that is the main sort of line I have been looking down. And hopefully we will be able to get some nice sort of examples out of this with some of the techniques that we have been developing in-house on how to really capture the uncertainty of these machine learning algorithms. And then hopefully we will try to apply for funding and really get this pushed out there and see the effectiveness of it I suppose.

Sean:                  That’s a fantastic kind of summary of what has been going on with the project and some of the kind of wider concerns. Yeah I mean obviously the ocean covers so much of the planet’s surface, I mean yeah I can see how that doesn’t scale massively well.

                            Is there any kind of thoughts of kind of alerting whales to the presence of the boats or is that not, like watch out kind of whale noises?

Calum:               It’s interesting because every time I have this discussion with essentially AI software engineers, technology people and myself included, the first thing that we think of is let’s do an active approach and sort of tell the whales that we are nearby, use the sonar. I saw this really interesting sort of piece online about using reinforcement learning algorithms to essentially learn the patterns of oh this is how you communicate so if we learn to communicate back this is going to be great. And to be clear that is a small research field out there. 

                            The problem is these devices that need to be used are actually going to end up screaming and yelling at the whales essentially and it actually physically damages their ear drums. So we are actually putting more noise pollution in there. 

                            And what’s really fascinating, at least in the UK, is you need a licence to be able to have these, they are called Active Acoustic Deterrent devices. And like I said you need a licence for them and they are very, very rare to give them out. The only time they will really actually provide them for, at least the UK waters, is if you are wanting to build some sort of structure and you need to dispel all the marine life there.

                            There is actually also some evidence showing that other marine life such as dolphins will actually tune out these sort of active deterrent devices basically as white noise and ignore them. So we have added more noise pollution and ended up damaging whales so that they can’t even hear other ships. We are also going to be increasing the stress for their feeding cycles meaning they are less likely to breed so that birth death ratio is even worse than it was before.

                            So there is some research but it’s highly discouraged as it turns out by the marine biologist side of things.

Sean:                  And yeah completely rightly we shouldn’t necessarily be telling them to get out of our way anyway particularly when we are in their environment. 

                            I mean obviously they are both under this overarching kind of umbrella of the environment that this episode is about but they are extremely different ends of it aren’t they? I mean effectively helping the farmer effectively as a business versus kind of saving the whales, which you know, pardon the pun kind of thing.

                            But just going back to the farmers briefly Karl. I am just thinking this is just an evolution of technology, farmers have been embracing technology from say learning about barometers to combine harvesters that drive themselves effectively or certainly guide themselves. So you have got these sensors, you have got some algorithms going on that are kind of trying to ascertain that you know, that the sensors are either, if there is a fault you know about that. But generally speaking you are getting data  that’s fairly accurate. But you are not kind of making automated decisions to say water as  a result or whatever you are still keeping the farmer in the loop.

Karl:                   Yeah. We are still keeping the farmer in the loop. So this data could be used for all sorts of purposes such as automating the watering or telling the farmer the optimal value of fertiliser to use. Because fertiliser in itself is damaging to the environment so it releases a lot of nitrous oxide, so by optimising fertiliser for example we can even help our carbon footprint or the farmer’s carbon footprint. And by reducing the wastage and things like that you can probably have some benefits for the consumer with better crops, better yield and low prices in the shops for your produce and things like that.

                            But I guess there is an explosion in IoT technology and that is always going to be the case. And one of the most important factors here is privacy. So like I mentioned earlier, in the wrong hands this data could be used for catastrophic things, damaging farms, damaging crops and this is food to provide for a population. So damaging-

Sean:                  So food security basically, yeah, yeah.

Karl:                   Yeah so food security that’s the word yeah. Food security. So there is a few different aspects to it so yeah. So most of it is like about the data, can you trust the data and what can we do about that to ensure that we can trust the data and how can we preserve confidentiality? So you have got a nice smart farm and everything is working well but how can we keep that information to ourselves if you like so that others can’t steal?

                            So you are monitoring whales so from what I have read you are using computer vision to detect the whales is that-?

Calum:               Yeah so there is a couple of ways of looking at it. So, and again it’s kind of like the different strands that we have been thinking about it, so one of them is using computer vision on the like say for example drones being able to go out and spot the whales. And even then that’s kind of interesting because you find that you mostly try to look for the blows that they do when they go up to breathe. And again it’s interesting that it is a very specific sort of things that they are looking for.

                            And then the other way that they use I guess computer vision, is they use it on the spectrograms that comes out of these hydrophones. But again it’s kind of interesting because I think Google did a big push towards this and they have like eight terabytes of data and they are able to get these really accurate sort of CNNs that really were great. But I had a workshop, like everyone was kind of quoting, saying there has got to be a better way of doing it but we are not quite sure how yet.

[00:20:21]

                            Anyway sorry, I am not sure if I answered your question other than just rambled on.

Karl:                   No that’s fine. So I was just thinking about computer vision, how can we use that computer vision with this project in AgriTrust? So I felt like maybe mounting a camera to monitor like livestock protection for example for prey or to monitor the behaviour of like small ecosystems, populations of different animals and things like that.

Calum:               Yeah actually that was the sort of thing I was sort of intrigued about as well because we are currently starting another project. I won’t go too much into it because it does fall into this realm where they kind of want to use drones to inspect the health of trees. And I am sort of intrigued like in the world of agriculture is that beneficial to have I guess drones or robots to investigate I suppose using computer vision rather than a human from a, I guess a safety point of view? 

                            And does that ever occur that it is not safe for the human to actually inspect the crops or would they actually end up damaging the crops themselves?

Karl:                   It’s the thing isn’t it, especially when the farmer has got a lot of moving equipment on it as well, you have to be careful with the drones.

Sean:                  I don’t know if you are aware of the hands free farm where they are using smaller lighter tractors rather than the huge kind of one tractor fits all possibilities. And the hands free farm maybe you would choose five drone tractors that do the same or more amount of work than the one single tractor but of course you are kind of spreading the weight. 

                            You are also having a little bit of redundancy so if one of them goes wrong you have got the ability to go, well four will still do the work but it might just take a little bit longer. Rather than your one tractor has a problem and it’s kind of, you know, it’s game over for a while.

                            So I think there is lots of stuff going on in that space isn’t there?  I think it’s really interesting.

Karl:                   Yeah there is a lot going on but it’s like again it all comes down to trust again doesn’t it? Trust in this autonomous farm to go and let it do its thing, how can you trust all these tractors to go out and-?

Sean:                  And even just a simple thing of okay there is a right of way footpath across the farm and what does the tractor that’s autonomous do when the rambler is exercising their right of way, you know, does it stop, does it? Well hopefully it stops at the very least b but how does that communication happen? I don’t know. I don’t know.

                            I mean we are slightly off the track, off the farm track.

Calum:               But I guess this does raise an interesting point about autonomous systems in general being deployed to be trusted by the public. But not even just in a case of safety but also am I actually going to be saving money by deploying this autonomous system. 

                            I am just sort of curious with the agriculture work that you have been doing, is the stakeholders, are they wanting sort of like hard numbers of improvement in crop yield or are they wanting the crop yield to be the same but improvement in another aspect? That is something I am intrigued about.

Karl:                   Yeah that’s the thing. So a few of the farmers feel like they want to improve the crop yields but at the same time a lot of them are interested in reducing their impact on the environment as well. So by optimising the fertiliser and things like that they can, they can do that but people are interested in it for a lot of different reasons really. But yeah improving the crop yield and reducing their carbon footprint is some of the biggest reasons why they want to adopt this technology. And also they can probably chill out for a little bit can’t they then?

Sean:                  Absolutely. If you are not walking down to the top field or up to the top field to check your soil acidity and you have got a sensor reporting back, absolutely.

Calum:               How many, sorry, sorry this is just a weird little technical question I just had. So how many times can the robot test the soil before it needs to see, I don’t know, like I mean this tells you how much I know about agriculture, but like almost wash off I guess the sensor right? Because I am intrigued  like how much acidity or alkaline concentration is still on the sort of prodder or so to speak?

Karl:                   Yeah so these are static boxes that we’ve got, so little boxes, and you can’t see me on the podcast, but they are only small. And these pH sensors, so they are buried in the soil and about every hour they get an update or the farm gets an update of the soil.

                            But one thing about these soil sensors is that they need some moisture to get a good result on them, these electrical type.

Sean:                  So are they testing like current between two points in the soil?

Karl:                   Yeah that’s right, electrical conductivity yeah. So they pass a current through and based on the resistivity or the conductivity they can infer the pH of the soil. So every hour they will get an update from these sensors.

Sean:                  There is a thing, we were just mentioning it kind of in terms of the agriculture and the trust side of that, what about during kind of your looking at marine life populations and things in PREVAIL, what are the trust kind of areas there? Because not to say about the people in Santa Barbara and the hydrophones and all that side of it, but they had no reports of whale problems. 

                            But as we said sort of slightly earlier that you don’t always know it’s happened right? The best way of working it out is finding unfortunately a washed up beast that has had some run ins, I mean how do we improve that side of it?

Calum:               So there was an initiative launched by the International Whaling Commission IWC. And to try and get a better idea of the current problem they set up a database where you know they can, you know captains and maritime operators, almost anybody can report whale sightings. But more so actually if they have actually collided with a whale and if this has ended up with a death or serious injury. But I am going to go back to that horrifying word again, this is all voluntary.

                            So even though it’s in that database, that’s the issue there is, and to be honest with you like maybe, I don’t want to sound too down and grim about this, or maybe I will. I will start off by going a bit down and grim and maybe we can end it more positive about it.

                            So you have these, the thing that we were definitely thinking about right from the get-go is no matter what it is that we develop or try to introduce into the maritime domain, at the end of the day there is no legal reason why they should use it or they should follow it, right. Because money is a big factor in this, so that is something else that we have had to consider is how do we advertise this in a financial way?

                            And there is a really nice paper out there that has been a really good study into this. Finding out that maybe one in every ten collisions results in significant repairs needing to be done which can cost hundreds and thousands of dollars, which also includes the time that the ship is inactive and not able to bring in income right. So that is one way of looking at it that we are going to try and push this forward and sell it.

                            But the other thing is it turns out a lot of people are wanting this to actually be a solution right. So they did, back to the Hellenic Trench in Greece, they interviewed ferry passengers and asked them if we were to go at a slower pace which results in us colliding with less whales or even if we do collide with less whales there is a significantly reduced chance of a fatality, would you be okay with this? And seventy-five per cent, I think it was seventy-five, seventy-five to eighty per cent of them went yes absolutely. So there is this also PR pressure as well that is really starting to come about from this. 

                            And from the discussions I have had with a few others who have been looking a bit more into the survey side of things are finding that this is more and more as, and I am sure that Karl here will agree, is this idea that there is a bigger pressure from I hope society in general about everything environmental. And how can we solve it with AI and autonomous systems but likewise I think society has been quite good at being quite cautious and questioning of this.

                            So it is really interesting to see that society is wanting us to be a bit more responsible but for a solution to present itself and to be adopted if actually the evidence is there to be able to trust and take its advice seriously.

Sean:                  And did the other thing,  not to kind of go on about this, but I was just thinking again about those hydrophones. I mean is there a potential solution to have ships equipped with basically whale detectors?

Calum:               Again yes but it does have some caveats to it. So the first caveat is what they  used to do with these hydrophones is they used to have them towed along the back of the ship, they were particularly large. But what you find is depending on the size of the ship, you will incur some sort of sound pollution and you are trying to minimise that so you need to have some distance away. So this could be almost kilometres long depending on the size of the ship. And then because of that size you need to slow down the ship so that they don’t entangle themselves. 

                            So there was this question for a while of if we just tell people to employ hydrophones to look at whales and if they do hear them to slow down. But if they are already needing to slow down does that actually cancel out the problem in a weird way? So we were thinking about developing some sort of idea of using other forms of information about when to deploy these hydrophones to get a better understanding was one way of going about it.

[00:30:00]

                            But I have been having discussions with TALs regarding this, again very informally I would like to inform the listeners, because there is some internal discussions about actually how useful this is. And again this was just being all informal with a colleague of mine. So what it is is they are thinking about having these hydrophones more placed on the centre of the ship, on the hull, so you know because the problem is you can’t have it at the back because you hear the wrong way right. If you have it on the front though you have got too much of the current and the waves that will cancel it out.

                            So they are finding a nice way to place it on the hull. And this way you still need to reduce the speed of the ship a little bit more for the sound but not to the same detrimental effect as it were if you were towing an array of them. So we have kind of been working off that sort of hypothesis at the moment and we are hoping to get a bit more of a clearer understanding of these exact numbers. I think the estimate was about maybe seventeen or eighteen knots I have been hearing. 

                            And we are kind of really looking into that a bit more and trying to use more of these realistic numbers in the models that we have been developing to kind of get a better understanding for the feasibility study. And make it a bit more compelling when we go can you please give us some more money so that we can maybe actually use this on a real ship.

                            Because to TALs absolute credit they have actually been very excited about maybe integrating this with their own ships to actually see what is happening. We of course just need the money and the next step of this project is to really show that this does have legs and then take it out to the real world.

Sean:                  Karl I will come back over to you in a moment, this is one thing I was thinking. Because as you may or may not know this but I am a videographer I have had to deal with sound problems quite a lot. So I suspect it is a similar thing to getting wind on a microphone and there is some incredible neuro network techniques these days to get rid of those things in terms of recordings. And I wonder if perhaps some of that needs to be applied to the kind of hydrophone technique but anyway that is perhaps a little bit speculative.

Calum:               No I think you are absolutely correct there in saying. And we have, or rather I have, been looking at this convulsion network that has come out of Google along with the open dataset. And we are trying to look into that a bit more about injecting some verification methods on how to really synthesise these controllers that really help trade up all these different the speed, the safety of the passengers on board. Because there is actually risks to them as well unfortunately as well as their costs that might be incurred via crashing with the whales.

Karl:                   I have a question Calum about the security of the models. So for example what is stopping poachers for example getting their hands on these models, flipping it the other way if you know what I mean like?

Calum:               No absolutely. I think the thing we are kind of relying, well not relying on so to speak, is whaling is illegal in a good amount of the waters here and there. It’s mostly just a practised, I don’t know, export is the wrong word, but it is a practiced activity I suppose in smaller parts.

                            And again it’s kind of weird to think about what do you do in those sorts of situations? Because again going back to the start of this mini discussion we have been having is there is again no legal requirement for anybody to even use the sort of technology. Let alone use it for something that’s a bit more I guess nefarious right in regards to this particular proposal.

                            But again I think there is pushback from society in general on the environmental thing which would encompass this as well is our understanding and our feeling.

Sean:                  Karl and Calum it’s been amazing to have you join us today and talk about the two different projects, very different projects but obviously both very heavily environment related. So it just remains to say thank you, thanks Karl.

Karl:                   Thanks very much for having me, it’s been great.

Sean:                  And thanks Calum.

Calum:               Thank you.

Sean:                  If you want to get in touch with us here at the Living with AI podcast you can visit the TAS website at www.TAS.ac.uk where you can also find out more about the Trustworthy Autonomous Systems hub. The living with AI podcast is a production of the Trustworthy Autonomous Systems hub. Audio engineering was by Boardie Limited. Our theme music is Weekend in Tatooine by Unicorn Heads and it was presented by me, Sean Riley.

[00:34:20]