I have recently come across a user (I am sure there are many) that is uploading images that are AI generated or at least heavily enhanced. I realize this is probably nothing new, but I feel that if this behavior is allowed to continue this will no longer be a nature lovers’ site. With the rapid increase in AI technology and mainstream access to AI generating tools, it seems that it should be required of users that they specify if their images are AI generated or enhanced. Not divulging the information that images have been altered in a way that would alter proper research, should void a user’s profile. How can this site continue to be for valid research if the “observations” aren’t valid? We are bombarded with false images, false videos, and general false information nearly every where else on the internet. I came to iNAT to upload and view authentic observations and I donate for that reason as well. I humbly ask that someone please intervene. I also humbly ask that all users actually LOOK at ALL of the photos of an observation to verify that the images are real. If images appear to be taken in a professional studio, cross check that with the location of the observation. Thank you!
This issue has been discussed in a couple of earlier threads:
- https://forum.inaturalist.org/t/ai-images-and-how-to-flag-them/58420
- https://forum.inaturalist.org/t/beware-ai-images-on-inat/44346
You may want to take a look at those threads to see other users’ thoughts on how to address the issue.
Yes, I would say anytime that you think an image is wholly AI-generated, you should flag it. Photo enhancing with AI is a tougher matter to handle, but it is probably best to start by politely asking the user whether they have used AI-based enhancing tools. There have been some cases where users have added photos that were so AI-enhanced they could no longer be IDed. One solution is to disagree to a higher level ID here if the photo has AI-enhanced attributes that combine elements of multiple taxa. But I too wish there were a clearer policy and way to deal with these situations.
If the picture were wholly or mostly from AI, it would seem reasonable to mark it “no evidence of organism.” Don’t know if we want to do that.
Thank you. I marked as no evidence last night. My issue with that is that I can’t honestly say there is no organism. It is possible that the user photographed the observation and then used AI enhancing. It is still as offensive to me. However, I know how sensitive this topic is becoming so I want to be clear. Are we saying with that “solution” that there is no evidence of organism in the photo due to the manipulation of the image? I can agree honestly with that. Are you able to tell me if this is only a temporary solution until the situation can be addressed more firmly? Just curious…
Maybe a feature request for a new kind of flag similar to copyright would be a good idea?
When AI is used even just to upscale images. It can often “fabricate evidence”. Especially with things like wing veins of insects. It can make up veins that don’t exist.
Seems like AI has already ruined iNaturalist then.
I just read the guidelines again and it is mute on the issue.
Maybe a photo pre-processing section in the guidelines would help:
Cropping should be encouraged.
I think adjusting light and white balance are acceptable.
Retouching should be banned or at least discouraged.
AI enhancement should be discouraged, keeping in mind that it might not be a conscious action just a device setting.
Observers might need advice which features would ruin their observation photos (focus stacking should be OK) and how to set up their device to turn those features off.
Most guidelines were decided upon 10 years ago I suppose, usage changes…
I just got a new phone and AI enhancement is turned on by default. I’d therefore say most pictures taken by a phone camera in the last few years are highly likely to be enhanced by an AI algorithm already.
There is no black and white, just life and death (maybe). :)
I would say that, simply, it would be better to strictly forbid the use of such images and users that upload such pictures should be seriously warned not to do so anymore.
There must be no space in iNat for what is not real.
NB: i am not referring to photo processing.
“AI enhancement is turned on by default”
That’s…pretty terrible. So now less realistic photos will become the default, all because of tech companies obsession with forcing AI into everything
It does depend on how much the enhancement is actually doing…
but, outside of iNat, people LIKE Glorious Technicolour, who cares if the ‘enhanced’ details are wrong. I mean, WHO will ever notice, a missing leg or an extra wing?
For iNat obs, please upload pictures of reality.
@reiner which phone is that if you don’t mind me asking?
Yes, I’m working on guidelines plus a new flag for completely fake content, plus a DQA vote for assessing the accuracy of the evidence provided for situations where gen AI or excessive cloning and stamping to “fix” photos. But it’s not easy! It’s a constantly shifting world, photographically, and everyone will have their opinions. FWIW, I’m looking to wildlife photography competition rules, like these, as a starting point.
There’s never going to be a silver bullet and I think we’ll have to strike a balance between accepting what tools are available to and used by most people and what’s acceptable and not acceptable when it comes to accuracy.
I got a low end phone with android 14 and upgraded to android 15 (moto g55, but I don’t think brand would matter so I initially wasn’t going to say).
Camera → Settings → Shot Optimization (Enhance photos with automatic tuning and AI scene detection) is on by default. I’m not sure what this would actually do but for sure it would make something up to make a photo look prettier (may not be significant in most instances).
We haven’t been ruined yet. Your concern of fabrications is exactly why I am asking for help.
Exactly. And I am requesting an update to the guidelines and for the guidelines to be firm. We just can’t depend on (some) people to do the right thing on their own.
I agree about forbidding AI images. “There is no space in iNAT for what is not real.”…exactly. But I was starting soft with my request. However, “seriously warned” is too soft in my opinion. 3 strikes and they’re out. I would actually prefer only 2 strikes before account deletion. Everyone should be allowed to enjoy themselves, but not at the expense of the whole.
Your observation about people liking that “glorious technicolor” is part of the reason I am here. It is the reason for the work I do on our land. I am striving to share REALITY with the world in any way I can. I have a dream that the more people who are exposed to nature (in all its GLORIOUS messiness) the better chance we have for a future where some nature still exists. My concern is that the younger generations will no longer be able to appreciate the realness of the natural world if sites like this one allow the overly “perfect” (in someone’s opinion) images to prevail. Nature doesn’t need a Technicolour revival. Thank you for noticing.
Thank you for the work you are doing. I really appreciate it. I don’t mean to sound snide (I make it very obvious when that is my intention) and I truly apologize if I sound rude with my following statements. Scientific research should be the driving force behind the guidelines. Then there is no need to worry about varying opinions on photography. Pretty pictures are nice. Real photos belong here. Thanks again.