Is there a way to opt out of AI identifications on iNaturalist.
Whilst I like to use the AI as a way to check on my own identifications and even to get suggestions for further research if I choose not to add the taxon suggested I would rather other people are not allowed to them come along and add what the algorithm tells them it is. If I wanted to know the AI’s ‘opinion’ and add it I’d have done so, so someone else doing it gives me no useful information on what the organism is that I’ve not already considered.
How do you know the IDer that comes along after you didn’t consult the CV as a suggestion but used their own knowledge to agree with your ID (the same approach you used)?
Even if the added ID has the “CV” symbol, it doesn’t mean the person who added it relied on the CV and didn’t have their own knowledge. Sometimes it just means that they couldn’t remember how to spell the ID and it was easier to have the CV pop up the options that they could select. (This is often my problem).
You can opt out of Community IDs if you want, and not accept anyone else’s opinions.
Yes. I use the CV Suggestion almost all the time because I’m lazy and don’t want to type in the name. And occasionally I want a second opinion even if I’m mostly sure of my ID. Or I just want to see how good the CV is in interpreting my photo.
Sometimes if I suspect an ID was CV based (e.g. a species-level identification of a species in a very cryptic group from a new user with the CV badge) I’ll ask the user if they could elaborate on their ID. I try and ask in a very friendly and non-confrontational way (“thanks for the ID! I find it very hard to tell speciesone from speciestwo and speciesthree. do you have any tips?”). often the ID will be withdrawn, sometimes it will turn out they’re an expert who recently signed up and then i get to learn how to distinguish those better!
OK if that’s the system so that’s what it is thanks.
I do know the some people will be using it just the save typing the name out, does mean I often have to just disregard what may have been a valid ID suggestion because I cannot tell it from a computer generated one.
I (still) don’t understand how one can not use the CV suggestions. “Long ago” when I tried typing in the ID (on my own, or others), the CV would automatically fill it in. I liked that to save on typing, but I was trying to avoid this very conversation from others. But, I gave up and now I always use the CV for the ID’s about which I am confident.
I always work via the website, often Chrome now, but sometimes (and in the past) Firefox. (when ID’ing for others I simply “agree” if they’ve chosen what I know is correct).
It depends. If one is using Identify, the CV suggestions are not automatically displayed in the field for entering an ID.
I also end up typing IDs a lot of the time regardless of the interface I am using, simply because the ID I need is often not among the suggestions.
You can ask the person who provided the ID. Sometimes their profile may also provide a clue about their background and whether they have some knowledge about the taxon they provided an ID for.
You are discarding data from busy identifiers (in particular scientists who are taxon specialists) who choose the most efficient workflow.
Typing the name out
letter
by
letter
is not going to happen.
I don’t think you have to disregard anything, at least most of the time. A quick visit to a user’s profile will often reveal how experienced they are with a taxon. You can look at their identifications to see if they seem to know their stuff. You can leave a polite comment asking for more information.
This is semantics, but it’s important semantics, I think: There are no computer-generated IDs.
The Computer Vision provides a list of suggestions, but it can’t actually pick one from the list and post that as an ID. Only human users can do that. (You’re not allowed to have an automated bot making IDs for you!) The actual choice is always made by a person. The CV is a tool that can help them to make that choice, but it doesn’t actually choose.
On iNat, there is always a risk of someone making an ID thoughtlessly, with or without using the CV. But they just get one “vote”, and they always can be outvoted by people identifying more carefully.
You are only responsible for your own ID, not for deciding if other people’s IDs are valid. You should only make an ID if you think it’s correct, regardless of how the other person made their ID. (If you know that their ID is wrong, then you can tag in other people to help, or just leave it and wait until other people show up.)
OK, so how would you propose changing the UI so that people make different choices? i.e., How would you discourage them from using the CV uncritically while still making the tool available for those who wish to consult it as a starting point for research (or who use it because it is quicker/easier than typing)?
My role as user is to provide the result of my experience with the app. It is not my place to direct the solution. The developers no doubt are users as well, and I trust my feedback helps them see it from a perspective that is less immersed and expert than their own. It’s a slippery slope to be an expert and fall into the trap of believing “there is a right way” of doing things. The goal is more, high quality inaturalist observations and especially encourage people so they take the time to stop and look in the first place.
And as a user, perhaps you have ideas about how you would want the UI to work differently so that it guides your interactions in a way that you feel is more helpful.
We are encouraged to keep forum posts constructive – it is generally more constructive to not merely suggest something isn’t working, but, when possible, to indicate what preconditions might enable these things to work better.
There have been many threads on the CV and the frustrations it causes for identifiers when people use it too uncritically. In other words, staff and developers have presumably already gotten this feedback multiple times. What is notable in this context is that nobody seems to have found a good solution. Various suggestions have been made and various possibilities discussed, over the course of years. And yet there still seems to be a lack of good ideas about how to effectively make this happen.
I asked what you would suggest as a way of getting new perspectives and moving the discussion forward beyond the diagnosis (which most of us are already aware of) that the UI encourages uncritical use of the CV, in order to consider what can be done about it.
