People Using (non-iNat) AI Identification on Observations

General
,
1

There was a recent incident on one of my observations. I don’t want to call out a specific person but I’d like to recount it:

I got a long comment from someone on one of my observations about how they thought it was one thing or another based on google AI suggestions. Then they deleted the comment and made a kingdom-level identification (I uploaded it with a different kingdom-level id so that itself isn’t strange). They commented on the id only with a link to a biology slideshow (seemingly from a college freshman), which includes a very brief description of the kingdom. The slideshow does not have pictures that are similar to the species, and the description of the kingdom has nothing that would help with id. Having an incorrect or poorly argued/researched id is one thing, but with the now-deleted admission of AI I can only think the only reason why such a resource, that neither supports id nor gives any kind of similar pictures, is included is due to AI suggestion, and most likely the whole ID suggestion in the first place is due to AI.

I don’t really know how to respond to this kind of thing. If there’s a similar topic in this forum I can’t tell (“AI” is too short a search term, and multi-word searches don’t come up with anything). Part of me wishes there was a process to flag the id or user, but it’s not like the user isn’t a real person. Someone who’s only made one id ever and only a few observations. I don’t like to discourage people from iNat. ….But I don’t want to encourage them making id’s that are clearly against policy to be able to independently verify the identification. This is less about this one user as they’ve only made one id ever, but AI is growing in use in society. I end up wondering how many new/infrequent iNat users are out there who don’t think too hard on the ethics of AI and why it’s important to have human verified-IDs, etc (Different from the case of people who might be rapidly making hundreds or thousands of AI IDs, who clearly are invested in iNat but for the wrong reasons)

4 Likes
2

There’s this current thread which seems to be discussing a similar issue (though in reverse, there the observer is using AI to ID and provide comments). If it’s close enough, moderators can add your above post to the conversation there:

https://forum.inaturalist.org/t/stance-on-ai-assistants-to-write-responses-to-identifications/77041

2 Likes
3

if you disagree with the person’s identification or would like more information on their thought process, you can always ask them in the comments on your observation or via a direct message.

4 Likes
4

And if that doesn’t help here is what I would do: duplicate the observation and then delete the first one.

6 Likes
5

Maybe I’ll do that. I don’t actually have enough knowledge to put the observation in any category, but I could ask them their thoughts, thank you

3 Likes
6

Large language models are notorious for citing sources that have little to do with the claim being made, so linking to a freshman biology slideshow that doesn’t back-up the ID does seem likely to be from an AI answer. Pisum’s suggestion to ask them their thought process will likely just result in further posts of AI slop. My suggestion would be to politely reply with a comment stating that the linked slideshow doesn’t seem to provide any clarification, then ping some active iNat IDers for whatever kingdom has been suggested and ask them to weigh in with their opinions. If the IDer is discouraged from posting such suggestions in the future, that’s probably not a bad thing.

7 Likes