Latest / Wildlife and Adventure Photography / The future of wildlife photography, AI and ethics

Transcript

The future of wildlife photography, AI and ethics · Wildlife and Adventure Photography

• 0:08
  > Graham>Foreign how does AI impact your photography right now? And stepping back from our own personal experience with AI, how does that impact in particular wildlife photography? Because that's the area that I'm most interested in, in what I do. But what are the pros and cons of AI as we move into certainly, a time where the influence of AI is much stronger than it used to be. And I would say that's likely to increase as we move forward. So that's the subject of this podcast.
• 0:52
  > Graham>It's just to kind of kick around some ideas. So what you think, please drop me, an email at Graham, photography. that'd be great. If you've got any thoughts or indeed anything else you would like me to cover on this or and on any other subject.
• 1:09
  > Graham>So just to start really obviously AI is becoming, I mean it's pretty much everywhere or that's the impression I get these days. And you might be surprised to learn that you've probably been using it in photography without really thinking about it too much. So if you think of some of the technologies we've got in things like smartphones, particularly the later ones, you're getting things like scene recognition, high dynamic range, exposure balancing, portrait mode where it's recognizing faces, maybe applying a bokeh or putting a background out of focus, handling low light, detecting faces, smiles, that kind of thing. All of that is handled by AI and this is built into the camera that you're using for dslr, mirrorless uses. There are some really cool autofocus and tracking modes. They are AI because what the system is doing, what AI is doing for the photographer is detecting eyes. It's predicting motion and setting focus or automatically tracking and focusing. So these things are really handy on the one hand, but also they are areas where you might be using AI already and not being particularly aware of it. And certainly something I use when I'm preparing photographs to go into my print store. So I need very high quality prints, too because I do sort of high end printing. I've been using AI tools for several years now and these are used to sharpen up images, remove noise, also expand images. So the software I use to blow an image up so that it can be printed, maybe a three, a four, sorry, a three, a two, a one, maybe even a zero. That's an intelligent expansion program. It's AI because it's pulling pixels apart and then deciding what needs to go in between those original pixels. So it's adding new AI generated pixels so that the overall image, it doesn't become, pulled apart, basically, which is what you'd get with a normal digital zoom. So all of this stuff has been out there a while, it will be there more and more. And how does that impact us as photographers? So, on the one hand, I would say it's pretty handy, to be honest, because it does make life easier. And certainly when I'm photographing wildlife, yes, there are skills that you need to develop as a photographer to understand depth of field, understand how you're going to capture certain situations. So, for example, porpoising dolphins, they are out of the water for less than a second. You can't possibly use autofocus to get them in focus and do a normal composition. You have to use different techniques.
• 4:11
  > Graham>So there is a skill involved in certain aspects of wildlife photography. But if you're on land and you are tracking an animal, perhaps running towards you, being able to set the AI in the camera to track is certainly very handy and it reduces workload. So when I'm teaching photography, one of the key aspects that I teach is understanding camera modes, because cameras are actually pretty technically advanced these days, certainly compared to the ones I started with. And if they can do some of the heavy lifting for you, it frees you up to concentrate more on the more creative aspects of photography. And obviously you still have the option to go fully manual on many cameras. So really how you use AI is up to you, but it's certainly there.
• 5:03
  > Graham>So what are the other things to think about? And the other big one for me is the ethical consideration. And, and as I was researching this podcast, what struck me very strongly is that the ethics of AI, I believe, are pretty similar to the ethics of post processing. And this is a conversation I've had many times over the years, with people who really want to know, particularly people who've used film, because for most of us using film, we didn't do much with it. When we hit the shutter button, that was it. That was the end of the story. It would be printed. People do some m nifty print work, but for most of us, that was it. That was the, the end of the creative process.
• 5:47
  > Graham>It was then just printing and obviously these days with post process, post processing, that is an integral part, again, in my opinion, of the creative process that we now have with digital photography. So if you've listened to me talk about the photographic, process, if you've been on any of my courses, I'll talk about visualizing the image. That's the first step, capturing the image, is the second step. That's what we do with the camera. Post processing is the third step. So that would be Lightroom or Photoshop or whatever you use. And what we're doing with that is bringing the image back to what we first saw. And again, I'll talk about how the way the camera captures an image produces a result that is not the same as the way our eyes and our brain capture an image.
• 6:42
  > Graham>If you think of it that way, sunsets are a classic example. You might have photographed an amazing sunset that really inspires you. And then you look at what's in the camera and it's pretty ordinary.
• 6:54
  > Graham>So this is where post processing pulls that captured image back to something that's more real, the real image that you saw. And then the fourth stage is simply where we let it go. It becomes art because other people will look at that image and apply the filter of their own life to what they're seeing. So what are the ethics around post processing and do they also apply to AI? Well, I would say they do. And for me, the key thing is, what's your intention? If your intention with post processing is to produce an image that most closely or more closely represents what you saw when you took the image, then I believe that is really the function of post processing. Unless, of course, you're doing something very arty, maybe, something that is clearly not meant to be reality, let's say, put it that way. And I believe the same things apply to AI. So if your intention is to deceive, to create an image that, ah, represents something that is not real, and you are perhaps doing that with some ulterior motive, whatever that might be, it might seem fairly innocent or it might not be, but to me that that's where the ethical problem lies. So that is really, the issue. So I'm hoping that makes sense. I'm going to dive through a little bit more in a bit more detail about, I suppose, the concerns and what we can do about AI. And it does. There is most definitely an overlap, I think, between AI and post processing. So whatever ethics you apply when you're doing post processing, I believe we should also be using those with AI. So the first of the concerns I want to talk about is just loss of authenticity. And in that we can lose trust with the people who are looking at the image. So in other words, it's much harder to know if an image is real or if it's been AI generated and to what extent. And obviously early AI generated images were A little bit comical, they weren't very good.
• 9:04
  > Graham>But, AI has really moved on fast. And the reality is that in the not too distant future, it will become almost impossible to pick the difference between a genuine image and an AI generated image.
• 9:19
  > Graham>So that is something, I think, a reality that we need to be aware of. and what does that impact? So if you don't really know if it's real or AI generated, how does that impact, competitions, journalism, conservation campaigns, you know, are they all going to get flooded with fake wildlife images? so that's definitely an issue. Another one is that it devalues the work of field photographers. So those of us who do spend a lot of time, getting images, does that devalue our work? Probably it does, you know, in that sense.
• 9:53
  > Graham>And, what can we do about it? So I'll come back to that later. But certainly one thing that occurred to me years ago actually when I was writing, when I first was living in Australia, I was taking, I was doing a lot of humpback whale photography.
• 10:07
  > Graham>And, and one thing I really had to clarify with some people is that when you go out in a boat, in the case, I was actually in, Western Australia at the time, but I was just making the point that when you go out, in my experience, I might see breaching whales during the, whale watching season, maybe one in three trips. It wasn't a case if I'd get out to sea and I've got whale, every time I've got whales breaching left, right, you know, off in the distance, close, all of that. And I'm struggling to pick which whale to photograph.
• 10:34
  > Graham>That's not the reality. Occasionally you'll get those situations where you've got multiple breaches going on, but they're pretty rare. So what's the expectation? And that expectations, this is going back to the early 2000s, was a result of wildlife programs at the time. And obviously when you look at nature programs, wildlife programs, they're showing interest in activity, of these animals because that's what gets people interested. But unless you watch the behind the scenes part of it, where they talk about how they spent months, maybe to get a single shot, just a few seconds of a particular animal, you can perhaps think that wildlife is more widespread than it really is, is more common than it really is, and that a lot of these activities are very easy to spot. So if you then go on a safari holiday, something like that, you can be very disappointed. So it's about setting expectations, but, but it is also about making the reality of what's going on, more accurate, because if you look at, what is it? The biomass of mammals on this planet. So if you were to lump all the mammals together that we have, and, look at that as one big lump, the percentage of that that is actually wildlife. So think about 5%, if you know better, please let me know. But the reality is that most of the, most the mammals on Earth are us and the animals that we eat because we protect them. If you start adding things like birds, and I think I'm right in saying chickens are the most successful species on Earth in terms of numbers of them and simply because they're our food. So the animals that are likely to die out last are the ones that we eat because we make sure that we create environments that allow them to survive and grow. They might not have a great quality of life, which is true of battery farms. But the point I'm making is that it's very easy to misrepresent how many of the kind of iconic species there are. So elephants, lions, tigers, all of that kind of thing, to misrepresent how many there are left in the world. And unless you understand that, it's very easy to think that a lot of the conservation programs are perhaps exaggerating the seriousness of the situations. This is where, we lose trust, we lose authenticity. And, also through AI, we can get misinformation, we can get misrepresentation.
• 13:08
  > Graham>So that will give false ideas about habitats, numbers, all that kind of things.
• 13:13
  > Graham>So, ethical treatment of animals is another one.
• 13:16
  > Graham>so this here, I'm thinking of staged wildlife encounters. So this can lead to stress and harm to animals. And if you go to popular wildlife parks these days, many of them, and I've witnessed this firsthand on several occasions, which is why I don't tend to go to these parks. But you'll have, an animal. And I'm thinking of a time in Namibia where there was a lion. We just found it early afternoon, male lion under a tree, having a sleep.
• 13:41
  > Graham>And they're about 12 to 15 cars around it and just waiting for something to happen. And again, there's this, wild animals get treated like performing animals. and that's a whole other subject. But, and this is because people's expectations are that they expect perhaps lions to be hunting all the time or doing something interesting. Whereas if you see them during the day, they're mostly sleeping. Now, there's another aspect to this which, also relates to the data that is recorded, with an Image and how that can be used by AI. And this is especially relevant for endangered animals and animals and birds that are subject to poaching. because it is possible to extract the gist, the global positioning, GPS data that went out of my head for a moment and perhaps identify exactly where the animal is. The thing that's more relevant with AI is AI, artificial intelligence's ability to identify an area from a photograph so it can identify specific locations that are in the background of photographs. I've seen this done in debunking, so propaganda stuff that some countries put out pretty regularly and claim certain things are happening. But AI can be used to identify not only the location but the time that that image was taken, by looking at various things in the image. But equally so that's good for debunking false news items and propaganda news items. but on the other hand that same capability can be used in a negative way. and supporting poachers. Another thing that a lot of conservation programs do is to collate data based on wildlife sightings. So again if this gets access, so there are issues around security here, obviously that information is accessed, then anybody who would want to use that for poaching or something in some other reason that is detrimental to animals, then that is going to be a problem.
• 16:03
  > Graham>there's also obviously copyright issues and creative ownership because if you're using AI to enhance an image, it will use images that are floating around that it has access to on the net elsewhere. So who owns those images? Are ah, your images being used by somebody else to create an AI generated image. So that's another one that is an area of concern and also as I've said conservation and science, AI is very useful for species identification and monitoring. But you can't totally rely on it because mistakes happen and you need to have real information. This is why so many, why main, the main conservation programs have people in the field actually out there looking at what's going on. and the other one, the other thing I want to mention with AI is just amount of energy it uses. AI systems are incredibly energy hungry. And it does mean that when you use AI you are drawing on resources. So this impacts sustainability, all this kind of thing. So while AI is very helpful and I've certainly used it, I'm also very conscious that when I do use it I like to have my conservation credentials and have them fairly, have them authentic and do my best to be as environmentally aware as I can. and part of that is knowing that when I use AI I am not being as environmentally, responsible, let's say, as I would like to be. So, you know, I'll own that. so it's, it's, you know, where do you trade that off? What's the thing you need to be aware of? So I'm going to end this. And there is a download which summarizes. It's just PDF, it's brief. it's 10 points basically that you can use to perhaps use AI in a more certainly a more aware, conscious way and in a way that is constructive. That it is not detrimental to not only wildlife but the environment in general. So I think the first thing to do, and this is also very ethical. I think this is getting the ethics right is a good starting point. So transparency and labeling, make sure that you label, AI generated or AI enhanced images. this is generally a requirement in competitions. and generally they're very specific on what you can do with it. So it's generally the kind of ethical post processing that I've spoken about is okay, but doing anything more than that is generally not okay. So it's important to be honest about that. This also applies to journalism. If you're involved in journalism and you're showing images as part of a story to make it clear where AI has been used with that image.
• 19:03
  > Graham>And also in conservation campaigns. So digital watermarks or metadata standards like C2PA, content, provenance and authenticity, they are very useful because it allows viewers to trace whether an image was created with AI or not. And again, it's creating that transparency so that you are being very honest about how that image has been generated. in terms of ethical guidelines and standards. If you go to photography societies, competitions, conservation organizations, they will mostly have coats of contacts, codes of conduct, I should say on AI. So for example, National Geographic Wildlife Photographer of the Year, they have very clear, as I say, codes of conduct on that. And it's, I think as a general practice it's good to conform to those codes of conduct. Okay. industry norms is another one. And what's acceptable, so that might be noise reduction versus image generation. So I've already stated that I use noise reduction, software. I use sharpening software and I use expansion software.
• 20:23
  > Graham>So the package I use is Gigapixel AI, to do the expansion. So what I'm doing in all of those is improving the quality of the printed image but without changing what's actually in the image. so I think that that to me is one one of the measures that I apply. education and media literacy. It's good to be able to teach audiences how to critically evaluate wildlife images. So again, being upfront about what's gone into creating that final image, and also be clear about the difference between authentically documented images and synthetic images. so again there are, there are things you can see in them that will point towards an AI generated image. But as I say that's getting harder and harder as AI improves. probably dataset isn't so relevant to this podcast on the jump over that this is more for people who are doing actual research and, and presenting results.
• 21:25
  > Graham>I would say, protecting sensitive information is really important. So I'm very careful to remove anything that GEO locates what I've shot. I'm also with the images that I publish, I'm also very careful to make sure it's really hard to identify exactly where those images are taken. what I mean by that, if you're in an area where it's just kind of trees and grass could be anywhere if it's a specific waterhole. So I belong to certain Facebook groups, one of them is Etosha in Namibia.
• 21:57
  > Graham>And I've shared more general images but at waterholes and I've noticed there are people there who are very good at spotting which waterhole you're at. They know the area very well. So just be aware of that. If you are going to post an image of an endangered animal or certainly one that is at risk of poaching, be very careful to remove the GEO identification, information to remove that better metadata. Ah, but also maybe try and crop so that it's not that obvious where that image was taken. So the, you know, the close you crop into the animal. That's why one way of doing that, I did make some notes about bias in AI and that's simply that AI will tend to pick images that are common on the Internet. So they tend to be more the I'm going to call it the tourist stuff but the stuff that people want to see. So that will tend to be the more iconic animals. It will tend to be those locations. that does mean though there's cultural bias in that the less commonly represented animals tend not to be, pop up so much. and also images that are taken more locally, perhaps by people who live in these areas, they don't get the same rating, let's say when it comes to AI searches as, the more tourist stuff because there's less of them. So this can also create a bias, a cultural bias. And whereas with what I do, I'm mostly concerned with the animals themselves, the backstory to all of that is where are they?
• 23:36
  > Graham>And because the biggest threat to most animals is humans and what we're doing, understanding the local culture and what's going on there, is also very important in understanding what the threats are to that particular population of whatever animals you're looking at. So the cultural aspect is important. It isn't something that I feature particularly in my images, but it will come out in backstory. So things like some of these podcasts, I've spoken about different areas and threats and all that kind of thing, that's where that will come out.
• 24:10
  > Graham>So it is important to understand the cultural aspects of a location because that will have a direct impact on what the local people are doing, to either support the local wildlife or, or threaten it because it will tend to be one or the other. Okay, I think, I mean there's a lot of things here that you can do. I've given some guidelines. There is the download that you can refer to. there's a link in the description which is just to give you something that you can refer to if you, if you want to do that. I think I'm just going to skip through those because I've touched on some of them. but how do you for example protect your work? So you could use metadata or watermarks to mark your photos as authentic. Certainly, watermark them as being, as belonging to you. as I said, height sensitive data respect wildlife. So I think a big thing is fake encounters. It's something that comes more and more. And another aspect of that is that what I've been seeing recently, in a few occasions now, this is in Africa, but I don't think it's limited to Africa. Are people having encounters with wildlife. And in fact the two most recent involved elephants. And the guides have got the people that they're guiding into dangerous situations because they're inexperienced. And on one occasion two ladies were killed by an elephant. And on another occasion somebody was certainly attacked. And I don't know the outcome. It was just something I saw on YouTube, a couple of days ago and I haven't got any further with that. But the bottom line is you really need to be working with experience guides and guides, ethical, programs that have, well trained guides, will have a training program for guides, people that I use, I know, know what they're doing. in fact, in Zimbabwe, which is one of the projects, that I've been involved with a couple of times, I want to get back involved when we're actually looking at doing something more formal. The guide there, Dean McGregor, is very experienced. He actually trains guys. He knows what he's doing. And I've said it in other podcasts, I have actually trusted him with my life on more than one occasion. And that is the reality of the situation. If you are going out with a guide and maybe you're on foot, even in a. Even in a vehicle, it can be dangerous. The other. So I think the two ladies were on foot who were killed.
• 26:53
  > Graham>And. And the other one I saw was in the Okavango Delta, where people are on these sort of punts.
• 26:58
  > Graham>And I've been in the delta on those things. You are very vulnerable. And, we were lucky. But when I was in one, I could see, hippos in the water.
• 27:08
  > Graham>They weren't too far from us, but they left us alone. But hippos can be very aggressive, very, very dangerous. This particular occasion was elephants, and it looked like they had young. And I'm guessing that that may have been the trigger for the elephants to attack. But there are certainly times when you need to be incredibly respectful of large animals, particularly because they will protect their young, or if they're in musk, they can be very unpredictable and very aggressive. So as somebody who is a guest in these lodges and other things like that, you are, trusting your life to guides, and if they're not properly trained, and this definitely goes on in Africa and, as I said, other places too, inexperienced guides are put in charge of people that they should never be put in charge of. And, the results can be extremely serious. So, I probably laboured that a little bit. But, you know, coming back to that, it's really important to respect wildlife. So I've sort of gone off track a little bit, but, here we are.
• 28:13
  > Graham>Okay, so, tell the story, tell the picture of your encounters. You know, if you're publishing things, give a little, summary of where you were, how you found them, what the experience was. Just to underline that this was an authentic encounter. I would say use AI sparingly. So there are definitely. And again, it comes back to the ethics that I've already spoken about. So I'm not going to go back to that. the usual Things, be aware of your footprints, try and keep your impact on the environment, the general environment, as low as possible. So use AI as minimal, as possible, or in a minute, minimal, possible way. And you know, stay human. Because I think the real power of these encounters, really, I think the power of any wildlife encounter, it's something I try to capture in my photography, is the connection with the animal. I won't do that. It's computer, it's an algorithm, it's people of software. the real power in my opinion of wildlife photography is connecting with the wildlife itself. It's having that experience of seeing another creature, something in its own environment which is different to ours. It has its own way of connecting with its environment, its own way of surviving, whether that would be feeding, avoiding predators, whatever it is. And that's sort of the backstory. And this to me is where it gets really interesting because it's beginning to understand how amazing a lot of these animals are. And so with my photography, my thing is I like to get in close and do portraits where I can. I will do contact shots as well so that you can see where the animal lives, maybe get a feel for the environment. Is it an environment that is very harsh? Is it an environment that is plentiful? You know, and all of these things are possible even in the same environment.
• 30:10
  > Graham>So I guess to finalize, I hope that's been useful as touching on AI. Maybe you weren't aware that you are already using AI in if you're using smartphones or even dslr, mirrorless cameras, you might not have, have understood that some of this is now artificial intelligence that's being used.
• 30:34
  > Graham>so really smart computer programs. And also I think it's, it's worth revisiting the ethics of wildlife photography anyway from time to time because whether it's what you're doing with post processing, how you're using AI and how you're using your images, I think it's good to always just step back and think about the ethics of it.
• 30:56
  > Graham>What are you trying to achieve? What's the purpose of modifying a photograph? And you know, what does that say about you as a photographer? And I guess it also talks about, you know, how are we as people, what are we doing as people?
• 31:14
  > Graham>All right, so, I hope I've earned a, like subscribe. I hope that's not got preachy. I try not to, but I do want to share what I feel are important issues and aspects of photography, especially wildlife and conservation. but obviously these things equally apply to portraits, people, life events, all that stuff. So that is it from me for this one. if there is something you would like me to cover in a podcast, please let me know. maybe you want to have a chat, maybe come on as a guest and share, your work, what you do and share your story. Because I think, how we all get into photography and what we do with our photography, I think there's a lot of interesting stories there. I featured a few already with, different guests on the podcast. So, if you would like to be one, let me know and, I will speak to you again in the, in the next podcast. Bye for now. Well, I hope you enjoyed that. Now, I just want to say thank you for tuning in and joining me in the Wildlife and Adventure Photography podcast.
• 32:18
  > Graham>If you have enjoyed today's episode, please give me a. Like a subscribe, maybe tell your friends, and by all means leave a comment. And if there is a subject you would like me to cover in the future, please let me know and I'll, be very happy to do my best. So thanks again for, joining me and I look forward to seeing you again next podcast.
• 32:39
  > Graham>Bye for now.