Experts warn increase in enhanced photos on birding platforms creating fake sightings, threatening credibility of tool used by scientists

For many birdwatchers, recording a species outside its normal range is the holy grail. In the UK, the discoveries often make national headlines. The western reef heron, for example, usually found in Africa and southern Europe, was spotted in a seaside town in north Wales in June, and widely celebrated on birding forums.

But a new scourge is threatening to disrupt the fun: AI slop.

Scientists are appealing to birders to limit their use of AI when editing images over fears that it could undermine the credibility of popular citizen science platforms such as iNaturalist and Macaulay Library, which are routinely used in scientific research to monitor species’ habitat range.

[…]

Lees points to the example of a false sighting of a red-winged blackbird in central Brazil, which is usually found in North America and had never been seen in this area of Brazil until the reported sighting. The bird was, in fact, an epaulet oriole – a common new world bird species – but the photographer had asked an AI platform to make the picture “look better”, which is when parts of the red-winged blackbird were added, resulting in the false sighting.

all 2 comments

sorted by: hot top controversial new old
[–] 11 points 3 months ago (1 child)

Bro just post the picture you actually took, that represents the experience you actually had. I don't really understand the way it feels like many relate to photos. It feels like many people want the photo to be better than the actual experience, rather than something that represents the experience even if the photo is flawed

  • source
  • hideshow 1 child comment
  • [–] [S] 1 point 3 months ago

    I can somehow relate to the issue. I am NOT a photographer, I can’t even adjust focus on non-phone cameras for a family picture. But when I made the yearly family photo calendar, Adobe Photoshop and Lightroom had easily accessible functions to make the calendar pop more (five to ten years ago). Now, I can see the attraction to ”clarifying” tools, but just as any other applications in society, we need to view this more critically and not just be amazed by how much more AI slop makes our work pop . We will always be amazed by something that seems like unreal or surreal.

    I naively never realised how this would expand to research beyond our shallow social media.

  • source
  • parent
  • 46 comments in c/technology@lemmy.world view that post
    [–] 143 points 3 months ago (27 children)

    Who the fuck bothers to post fake images on birding websites?

    The whole fucking point of birding is enjoying seeing the birds. Taking photos of birds is difficult and satisfying.

    Posting fake bird photos is fucking stupid.

  • source
  • hideshow 27 child comments
  • [–] 108 points 3 months ago (10 children)

    There's a weird level of brain rot going on with AI. My wife recently had some troubles trying to research who owned our home in the past (it was built in the 1880s); she posted to a local forum asking for some specific information about who & where they were buried and so on and was getting replies from somebody that were seemingly plausible but just had no basis in reality when she looked into it further. After pressing him several times for his sources the guy finally admitted it was ChatGPT.

    Just... why the fuck? Why would anybody do that? They wasted not only my wife's time but their own as well, they contributed nothing useful (and if anything, their "contribution" was purely negative).

    It's baffling behavior.

  • source
  • parent
  • hideshow 10 child comments
  • [–] 51 points 3 months ago (9 children)

    At least in that case they could be stupid enough to think ChatGPT magically knows everything (lord knows a lot of people seem to think that) and be genuinely trying to help. I can kind of forgive being incredibly stupid and not realizing it if your intentions are good.

    Posting fake bird photos requires making or pretending you made them. Either case is shitty.

  • source
  • parent
  • hideshow 9 child comments
  • [–] 15 points 3 months ago (1 child)

    The mental illness in that case is a photo hobbyist thinking, "I should have been able to take that picture; AI is just helping me."

    To me it is exactly related to "i should be able to answer that forum supporr question. AI is just helping me."

  • source
  • parent
  • hideshow 1 child comment
  • [–] 14 points 3 months ago (2 children)

    It doesn't. On a lot of phones you can just zoom in and the camera will use AI to fill in what it wants because of some "super zoom" (or similar) feature.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 7 points 3 months ago* (last edited 3 months ago) (1 child)

    That is in no way comparable.

    This isnt a Samsung moon monent lmao

  • source
  • parent
  • hideshow 1 child comment
  • [–] 0 points 3 months ago

    It's different, but very similar. AI will just fill in what it doesn't know with something that it does know.

    Really it's kind of worse than the moon photos. At least that is what the moon would actually look like.

  • source
  • parent
  • [–] 1 point 3 months ago*

    While it doesn't know everything like that, its gotten very hard to make the pro models fail on freelance ai training sites, they only fail on roundabout logic like "on the year the cardinals won the superbowl, the state that the kicker is from, had a protest that year, what is the name of the winner of the golden globes for best actor year the leader of that protest died." and it wouldnt fail at that probably.

    It's a chore to get it to fail on anything that could possibly have a source online (which is key here, you guys didn't have a valid online source so it hallucinated, when making it fail for training you need to provide an answer with sources to show the AI is failing)

  • source
  • parent
  • [–] 19 points 3 months ago* (3 children)

    Outright hoaxes remain rare and are often easy to spot, said Lees – nobody is falling for a toucan sighting in Siberia – but he says birders often use AI to edit and improve an image, and the algorithm can then introduce parts from different bird species to create the new image.

    My phone offers an "AI enhance" option after you take a photo. Mostly it tweaks the framing and colours, but it can also do some surprising things and produce a quite different image from the original. I played with it a bit when I first got the phone, and I saw it change people's eye colour, turn their head to face a different way, and add extra strings to musical instruments. So I don't use it, but it sounds like the problem is mostly people using AI image enhancers like that carelessly.

  • source
  • parent
  • hideshow 3 child comments
  • [–] 1 point 3 months ago (2 children)

    Google pixels AI enhance by default for the preloaded camera app...

  • source
  • parent
  • hideshow 2 child comments
  • [–] 1 point 3 months ago (1 child)

    Do they? I know they process the images, blending in takes from the different cameras, doing white balance and enhancing colours, etc., but the "AI Enhance" is a button you have to press. Unless they have changed that.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 11 points 3 months ago

    It's so ridiculous and unnecessary too. I've gone out and identified such beautiful birds, nature in general is so varied yet very often vibrant. It's a hobby that really does not need any alteration to what you see: birds are very pretty!

  • source
  • parent
  • [–] 5 points 3 months ago

    Pathological liars feel the need to lie and decieve unnecessarily and pointlessly at every twist and turn. AI is their new bread and butter. Ask them why they act the way they do, and they probably can't even tell you.

  • source
  • parent
  • [–] 4 points 3 months ago

    The very popular app ebird has leaderboards, that alone pushes lying and exaggeration. Obviously I am not saying everyone is cheating, just that some are. I would personally bin such fripperies.

    Feedback loops on more general wildlife social media also pushes a very different expectation of what is a good photograph and what is a hard bird to photograph/find.

    If you chasing clout from that audience then its putting ever more pressure on to get ever closer, higher framed shots of the most photogenic birds, AI (re)generation to chase that becomes inevitable for those without morals

  • source
  • parent
  • [–] 7 points 3 months ago (1 child)

    The internet peaked over a decade ago when bandwidth was alright for what we consumed and attention seekers have only infested mainstream media. You could navigate around them. Most forums have died since the early 2010s and you have to deal with these faking clowns even in the tidiest niches that we have left. People have become completely obsessed with attention and money. It‘s all the internet is good for anymore. At least if you ask most users.

  • source
  • parent
  • hideshow 1 child comment
  • load more comments (1 reply)
  • [–] 41 points 3 months ago (3 children)

    A lot of users here seem to be under the misconception that these are 100% AI generated images.

    The problem isn't really with that as the article states that this is an issue with using AI in the editing process:

    Outright hoaxes remain rare and are often easy to spot, said Lees – nobody is falling for a toucan sighting in Siberia – but he says birders often use AI to edit and improve an image, and the algorithm can then introduce parts from different bird species to create the new image.

    It's also worth noting that some phones allow you to zoom in beyond your camera's normal capabilities by using AI to enhance the picture. With some algorithms this might not be an issue, but if it's using a model that's been trained by some animal/object database then this can introduce innaccuracies that would cause problems for researchers trying to keep their database clean.

  • source
  • hideshow 3 child comments
  • [–] 21 points 3 months ago (2 children)

    This was a whole thing with Samsung phones replacing the moon with a fake one in their telephoto images... I hate how much processing is going into phone images now.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 11 points 3 months ago (1 child)

    You need a real camera if you want real pictures. The optical zoom on a phone camera is only a couple times magnification.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 5 points 3 months ago

    I actually did buy a Sony mirrorless camera several years ago when I was getting unhappy with the quality of my phone camera, right around the time the aggressive denoising tech was introduced. It made all the leaves in my forest photos turn into a smeary mess, not much different from an AI image.

    These photos were 6 years apart, but I bought the nice camera shortly after the first photo. Galaxy S10 vs Sony A9 straight off the camera, similar shutter speeds, but different weather.

  • source
  • parent
  • [–] 22 points 3 months ago

    I WANNA GET OFF OF MR PETER THIEL'S WILD RIDE!!!

  • source
  • [–] 13 points 3 months ago

    First we had Merlin Appholes who would submit misidentified and similar sounds as "sightings", but Cornell tightened the tolerances on the recognition model.

    Now this. But to be fair, even before AI there was a subset of "birders" who loved to exagerrate via ignorance. Cardinal = scarlet tanager (it's red), american robin = baltimore oriole (duh, orange), red-winged blackbird = Bobolink (it says this meadow has Bobolinks, therefore, any large bird on grass = bobolink), and so on.

  • source
  • [–] 13 points 3 months ago

    Science needs to depend on trustworthy observations. This is so not cool.

  • source
  • [–] 9 points 3 months ago

    the photographer had asked an AI platform to make the picture “look better”

    puffin filter on person with white hair

  • source
  • [–] 8 points 3 months ago

    I’m waiting for AI to delete the sum of internet accessible human knowledge.

  • source
  • [–] 7 points 3 months ago (1 child)

    "AI" birds aren't real!! /s

  • source
  • hideshow 1 child comment
  • [–] 6 points 3 months ago (4 children)

    I have used crowd collected data for research. You have to curate each picture carefully and cannot trust labels, there are many erroneous identifications.

    Maybe if you have access to pre-curated you can trust it a bit, but if you are a researcher and use images from these websites without curation you are already doing it wrong...

    AI altered images doesn't change things too much, it probably just increases the numbers of errors that already existed...

  • source
  • hideshow 4 child comments
  • [–] 7 points 3 months ago* (2 children)

    The “failure mode” of AI editing is different though.

    Humans (I guess) might mislabel something or take a bad shot. If they try to touch it up “traditionally” they could mess up the coloration at most.

    But with AI editing, now you have to watch out for fine details you’d normally use for identification being completely, convincingly fabricated, as the article points out, with altruistic intent from the user (who’s just trying to submit data that looks alright)

    The solution is global AI literacy; but that’s not going so well.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 3 points 3 months ago (1 child)

    For "AI editing", I don't have much of a problem if a model was trained to tweak the settings in something like Darktable to achieve a good baseline to start with. However, for contributing to research like this, I draw the line when models start generating their own pixels and overwriting the original image.

    Hopefully iNaturalist and other similar groups start to look at the metadata of submitted images to help warn/educate end users about this problem. That would at least help with the AI literacy issue. Those metadata tags are already being placed there by the most popularly used tools.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 4 points 3 months ago

    I'd argue that AI does change things:

    • on one hand it increases the percentage of invalid/unusable images. A mis-labeled image is still potentially useful, an image of a toucan in the Arctic isn't. This means that even large datasets become largely unusable because of the volume of fake images.
    • on the other hand any AI picture that slips through, taints the dataset and results in expensive manual cleanup to ensure data is reliable.
  • source
  • parent
  • [–] 3 points 3 months ago

    Lees points to the example of a false sighting of a red-winged blackbird in central Brazil, which is usually found in North America and had never been seen in this area of Brazil until the reported sighting. The bird was, in fact, an epaulet oriole – a common new world bird species – but the photographer had asked an AI platform to make the picture “look better”, which is when parts of the red-winged blackbird were added, resulting in the false sighting.

  • source
  • [–] 2 points 3 months ago

    This is going to be more problematic for less photographed birds.

    The more unique the bird, in terms of internet pictures, the more it's going to be changed with AI.

    Which means you're far more likely to see a common bird in the US or the EU appear somewhere else.

    Much less likely for AI to add some crazy unique rainforest bird to a park in LA.

  • source
  • 11 comments in c/technology@lemmy.zip view that post
    [–] 20 points 3 months ago* (10 children)

    Users can create high-quality fake images in seconds – or enhance a photograph by asking the AI to remove a branch or leaf that might be obscuring part of the subject, which can inadvertently introduce significant changes.

    Using AI even to edit an existing photo turns it into a piece of fiction; better to either accept a degree of imperfection or just take a few photos and pick one's favorite.

  • source
  • hideshow 10 child comments
  • [+] -7 points 3 months ago (9 children)

    Arguably human editing is the same, why even make a distinction?

  • source
  • parent
  • hideshow 9 child comments
  • [–] 7 points 3 months ago

    I appreciate when the photographer is forthright with it being an edit if they remove a branch/leaf/blade of grass, especially if they still share the original.

    Any post processing is altering the original, and anything AI can do to a photo was done by hand in the past, including adding or removing subjects. I agree with you it doesn't so much matter the method of alternation, it's still the person presenting the image that is responsible for its credibility.

    I also agree with the article that people uploading data to a crowdsourcing platform that is used by people trying to legitimately track animals to protect them is poisoning the well for people that are trying to do important work. A ton of resource is poorly funded, and having reliable volunteers is priceless. If people start throwing in made up datapoints, that takes a toll on real organisms. But then do you limit who can contribute, potentially excluding crucial photos? It just makes things a mess for everyone if people start becoming dishonest.

  • source
  • parent
  • [–] 3 points 3 months ago (1 child)

    Using AI for de-noising and upscaling is pretty normal these days.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 6 points 3 months ago

    And that's terrible. Imagine seeing a detail in a picture of a bird taken with off-the-shelf components and not being able to know whether that detail was hallucinated by denoising software. Unless that image is accompanied by the full weights of the specific AI version that did the operation as well as the random seed used for its stochastic elements, there is no way to double-check.

    There isn't even a guarantee that these modifications will be uniform. What's stopping an oh-so-helpful postprocessing software engineer from using AI vision to determine what the objects of the image are and then handing that information over to the upscaling algorithm to fill the detail with context-dependent hallucinations? Like, if the AI thinks it's looking at a bird, it'll fill the space with generic bird feather texture that might not fit to the bird in question at all.

    So you couldn't even come up with an ad-hoc postprocessing step that replaces AI slop with noise because the AI slop is trying camouflage itself as realistic. The only scientific option is to nuke all data about how the image looks below a certain level of detail, effectively blurring it until the AI slop is invisible. This is way worse than whatever sort of blurriness the AI is trying to correct.

    And using AI to redraw part of the image is even worse. You can't even blur the image to the point you're sure there are no hallucinations, the entire image is a fiction inspired by the image that hit the camera, which is now lost. Features at any scale could be made up, the result of the AI hallucinating something in place of an obstruction.

  • source
  • parent
  • [–] 5 points 3 months ago* (2 children)

    Because AI edits are far easier and quicker for most people than Photoshop, with the end result being worsened by slop generated to fill the space of cropped out elements.

  • source
  • parent
  • hideshow 2 child comments
  • [–] -3 points 3 months ago* (1 child)

    But Ai can also do it “correctly”, and humans can also do a terrible photoshop job, you need to be skilled to be able to use it correctly. It also has a host of automated correction tools, which is indistinguishable from Ai.

    Ai isn’t the problem with this issue, it’s the editing itself. This is just gonna make the rest of the issues come more to the forefront, they aren’t going to allow human editing and not Ai, that would just be going backwards.

    You say to accept imperfection, but a particular type of editing would be okay? That’s hypocrisy.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 6 points 3 months ago* (last edited 3 months ago)

    Human editing is a problem too, but AI is making such edits more commonplace and harder to notice. Any policy changes should therefore bar both types of editing, but AI edits are a far greater threat because one doesn't need to be skilled to make them.

  • source
  • parent
  • [–] 0 points 3 months ago (1 child)

    When a human edits a photo, there are artistic decisions being made, even for someone who's never used a photo editor before. When AI does it, it's slop. The human can attempt to recreate unique things they saw in the bird. AI makes up whatever without reason or thought.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 3 points 3 months ago

    Not that I disagree about AI outputting slop that doesn't match reality, but when the images are needed for scientific purposes, people shouldn't be trying to 'recreate' anything—whether themselves or with AI—but instead take photographs with that content there in the first place.

  • source
  • parent