The machine-generated content prohibition was designed to address the use of scripts and stuff to add IDs, comments, observations, etc., especially at a pace no human could match and without someone overseeing the addition of the content.
I think AI and images/sounds is more about whether or not the evidence presented is a real and accurate depiction of the encounter between the observer and the organism.
In an ideal world, all AI-generated content would be required to be labeled as such, but I think that is very unlikely to be enforced (or even enforceable for that matter).
I think on iNat generative AI could even be completely disallowed. Apart from the numerous ethical concerns there are regarding generative models and the companies and decisions behind them, this is another reason. People can do to their images whatever they want to make them more visually appealing, but iNat’s primary function is the recording and identification of organisms. That shouldn’t dissuade anyone who wants to to take and/or edit visually appealing photographs, but if the visual appeal comes at the cost of IDability, then that just kinda defeats the whole purpose of putting it onto iNat, does it not?
I do think copyright concerns are a large part of it as well, though. The issue is both, IMO. While big for-profit platforms may benefit from unregulated genAI–with its negative effects conveniently being hidden behind the huge dollar signs in the eyes of their share holders–places like iNat provide a chance to initiate a more reasonable stance towards them, so I really think there should be some clear statement soemwhere.
The whole image was fake and either AI or one of those crappy stock images. Someone already reported it as copyright, and I didn’t save it so I can’t show the image anymore. The “appear” part may have been poor wording, but I was referring to bad, non-photograph stock images and other similar media.
I agree that a flag similar to the copyright flag would be the best way to deal with wholly AI-generated content. I actually think the current wording of the copyright flag supports this as it says:
Violates copyright law or was created by someone other than the observer and lacks attribution (emphasis mine)
To my mind, a wholly AI-created image meets that second criterion - it was not created by the observer. The observer is also generally claiming a license over an image they didn’t create (which falls afoul of some AI image generation programs).
This approach might not cover a situation where an observer cites the specific AI that created the image and/or doesn’t claim a license over the content, but those are really rare occurrences, as it defeats the purpose of posting AI images for most posters I would think. The existing Copyright flag language could potentially just be modified to make it clear that it includes AI-created images.
One could argue that the person made the image by submitting the prompt, though. Exactly how AI images interact with copyright is completely undetermined at this point.
If I get an image by googling (submitting a search prompt), that doesn’t make me its creator (whether AI made the image or not).
If I commission an artist to make an image, and give them a prompt describing what I want the image to look like, it doesn’t make me the creator of the image.
The statement that
is incorrect.
The US currently does have rules to deal with some aspects of AI-generated content, as do many countries. Currently in the US, “works created solely by AI — even if produced from a text prompt written by a human — are not protected by copyright.”
https://www.theartnewspaper.com/2023/05/04/us-copyright-office-artificial-intelligence-art-regulation
https://builtin.com/artificial-intelligence/ai-copyright
Of course those rules may change, but they certainly exist. The current US rules at least don’t recognize the “prompter” as a creator who can enforce rights over the resulting image itself.
In that first article, it says “[W]hat matters is the extent to which the human had creative control over the work’s expression and ‘actually formed’ the traditional elements of authorship.”.
That still makes a lot of AI-generated images feel like a gray area to be honest. I feel like there’s a gradient of how much creative control can be in different AI-generated images depending on various factors, doesn’t it? I’m not a lawyer though.
Regardless, if they are not protected by copyright, then they aren’t really copyright infringement, since you can’t really infringe on a nonexistent copyright.
I hope iNat is clearly state that any observations which are generative by “generative ai” is strongly prohibited in any form. also gen ai is definitely violating copyright because these are stealing copyright materials without permissions.