Hello. I’m new to iNaturalist and I have no required knowledge to accurately id the living beings I find, so here’s my question: is it better to trust the app with the id (even though I cannot verify it by myself) or should I pick a very broad, but safe, id, like “insecta” or “arachnida”?
5 Likes
I recommend broadly identifying where you are sure, but perhaps try a little lower than “insect” and occasionally try to dig into IDs to try to get at least order level with insects. Otherwise you might end up waiting a while for someone else to find the observation.
I definitely don’t recommend just trusting what the AI says, as it ends up being wrong and it is harder to fix an incorrect ID over a broad ID (especially if the user becomes inactive later)
20 Likes
Welcome to iNaturalist!
One of the joys of iNaturalist is that people are so generous with their knowledge. I learn so much from others’ responses. Now I can generally get my IDs (outside of my areas of specialty) down at least to Order, and frequently to Family - and occasionally (joy of joys) to species.
I absolutely agree with @malisaspring - don’t go beyond your own knowledge. The AI is very accurate on some common species, and totally inaccurate on others, especially less common species.
I use the AI to get starting ideas for IDs, and by reading up on them I expand my knowledge. Then when submitting I back up the ID level to what I’m confident of. I’ll frequently put my guesses in the Notes section.
I started out specializing in plants and moths, so before I came to iNat had already learned about how difficult some IDs can be. Many moths (and, as I have learned, many insects) are not identifiable to species without dissection. Plants are easier (for me), but there are still many very similar species, identifiable by technical characteristics.
One ID tip - the “About” pages have, in addition to the Map, About, etc. tabs, a Similar Species tab. It can be very useful.
I hope you enjoy iNat!
10 Likes
Here’s an alternative view. There’s an advantage to accepting the AI name. As an identifier, I spend a lot of my ID time on one common genus of plants. If the AI says a plant is in that genus, I’ll see it and render my opinion. If the person is cautious and just records “plants,” I won’t see it and probably no one else will, either. So there’s a rationale for accepting the AI suggestion, as your ID may be seen by an “expert” more readily.
Overall I wouldn’t stress too much about which strategy to use. There are many people doing it each way, and each has advantages and disadvantages. (I do agree that using the AI suggestion and working to verify or refute it is a great way to learn and improve IDs as well.)
25 Likes
Welcome to iNaturalist!
The advice to post to a broad identification, or post to a narrower level, or to trust the AI as far as genus are all winning strategies to utilize as your personal knowledge grows. The community who will be reviewing your posts will guide you into more specific ID’s without lecturing you. We have ALL made mistakes in identification in our personal journeys to understand the natural world. The ability to post a clear diagnostic photograph is greatly appreciated. With participation in iNaturalist comes that motivation to develop better photographic skills, as well!
Enjoy the international community of naturalists!
7 Likes
Maybe it’s also a time to learn those common species known by AI, in one day you can lern hundreds of taxa because they’re unique, AI distinguishes them well for a reason. Insecta is a fine id, we will check it.
3 Likes
A lot of this depends on what Taxon you wish to focus on. The AI can certainly guide you to a ‘higher level’ ID, but for many Taxa it is not great for species. I’m a moth person (although not an expert) and rarely use it for Noctuid species. If you are new to non-human life in general, I would recommend becoming familiar with the predominant Life in your area, and then whittle it down to what you are most interested in. I have nothing against plants, but they just don’t attract me at all. There is no sea life where I live, so any ID’s I might give would be at a higher level. I grew up with insects, so that is what I know and like.
One of the great things about iNat is if the AI throws up an ID of say ‘Molluska’ you can access references on the internet, and refine the ID if you want. It is really a great way to learn!
Bottom line is that a rough ID is more likely to attract the attention of people who know that group. “Unknowns” tend to get lost, so if you use the AI, you are more likely to get a more refined ID. @janetwright explained it very well.
Don’t forget that image quality can affect the specific ID. I have seen some very worn moths, and some with purple tinge. It’s challenging for those of us who identify! And yes, ask questions. It may only result in a link, but it’s another opportunity to learn. Most folks are happy to help. -
8 Likes
Make a broad ID, definitely. We get a lot of new users who blindly trust the CV (computer vision) IDs regardless of whether they actually know the species, and it can be quite bothersome for identifiers. Please stick to something you can confirm. If you continue to use iNat you might find yourself learning some new species, so good luck!
3 Likes
A few pieces of advice, in my opinion:
- It’s okay for your ID to be wrong or unspecific. Someone might come along and fix it. But obviously you shouldn’t try to be wrong. You should make what’s your reasonable best guess. Also you don’t necessarily need to fix your ID if other people give a different ID, but whether you maintain a maverick ID should be based on a reasonable consideration of the evidence. The rest of you reading this can correct me if I’m wrong, but I think it’s fine to ask other people how they ID’d something; they’ll generally tell you some identifying characteristics for the taxon.
- You don’t need to give a species ID. Species ID is cool, yes, but sometimes you just can’t. Not even experts necessarily can sometimes. So it’s fine to just say that something is a “palm” or “beetle” or even just “bird” or “mammal”. (You may think that mammals would be at least easy to distinguish into different groups, but the “mammal” tag turns out to be useful for IDing scat.)
- Try to get as much information on an organism as you can reasonably get. Clues to an organism’s ID include: appearance of various parts (e.g. for plants: leaf color/shape/size and how they grow off the stem, shape of the stem, size of the plant, color/pattern/size/position of the flower, etc.), location (lat/long) and setting (urban, suburban, forest, grassland, wetland, etc.), time of year, time of day, animal behavior, and more. Take pictures from different angles if possible. Sound recordings are useful for some things. Add notes to your observation to describe how things act. And so on.
- When leaving an ID, especially one you’re not sure of, you can comment on how you came about it.
For example, if you have third-party information (e.g. “the park’s naturalist told me it’s this”) you can note that there. FYI, if you pick any of iNat’s AI suggestions, there’ll be a little icon on your ID that tells people it’s an AI suggestion, so they know where it came from. (That doesn’t mean it’s wrong or right!) - When you have multiple pictures, you can get iNat to apply AI to different pictures to try generate an ID. Some pictures may be more useful than others! iNat’s suggestions are based on the first photo if you’re using the website, so you may need to reorder the photos (Edit → Re-order photos) in order to try different first photos for the ID, but if you’re using the iNat app you can click “what did you see?” and swap between different photos for the AI to try each.
10 Likes
In my opinion, do not trust the AI, but still use it. You can either choose genus level, or, even better, check the suggestions and choose family or order level (you will need to do searching for that). Two worst things you could do are: 1) choose AI suggested species level without having even basic knowledge in the organism group; 2) leave the observation in Unknowns (it might get swamped there forever). Another not great thing to do is to add a very high (phylum) ID level, especially in the cases of plants and animals. Sometimes it cannot be avoided, but try to use it as little as possible.Then, if your observation received an ID, do not agree immediately, check the species and see if you are able to ID it after reading about. If you cannot, just do nothing if the ID has improved your original. In case the new ID contradicts your original ID, it is always wise to assume that the IDer knows better and to withdraw your original ID. Nothing of the higher said is obligatory here, though. You may see yourself which course is the best.
3 Likes
Yeah, don’t just agree with an ID because someone else said it’s a certain thing.
If you do, that counts as your also saying it’s that thing. So only do that if you personally want to say that it’s that thing.
If you’re not sure, do some research yourself. Maybe search for guides for identifying that species. Or check out all the photos of species in its genus or family (o of the genus or family itself). (Some species are the sole members of their genus!) Check out the various iNat suggestions; the AI probably thinks they look similar.
If you don’t have time, that’s okay too. But I’d say it’s probably useful to give broader IDs when you just don’t have good information. You can also change IDs in post – I often take observations and ignore IDs in the field, then ID and research stuff later when I get home. Open question to @jurga_li: when you say to avoid very high ID levels, what are examples of situations where you think this should be avoided? (I don’t mean to argue with you.)
2 Likes
Welcome to iNaturalist, and thank you for asking! You might find the following forum topic helpful, and it already echoes many of the great responses you have already gotten here:
https://forum.inaturalist.org/t/identification-etiquette-on-inaturalist-wiki/1503
4 Likes
Lots of useful advice here. Your idea to use a safe identification–the most specific safe identification you are able to–is fine and very helpful for the people you want to given a more refined identification. Also, you might learn a lot by diving into more specific possibilities for identification based on iNat resources (like the map, descriptions, and photos). This can also aid identifiers, even if your ID is not accurate. Welcome to the Forum and to iNat!
2 Likes
I definitely recommend using broad IDs. Like if it’s something looks like a butterfly, but you have no clue how to identify butterflies or moths, do Lepidoptera. Or if you think something is a jumping spider, but you aren’t sure, do Araneae and in the description say “I think it’s a jumping spider, but I’m not sure.”
3 Likes
Definitely don’t just trust the ID to species level if you are unsure, because sometimes you will just get someone agreeing with your ID who is equally unsure, but trusts you as an expert. It’s better to put a generic ID if you are unsure. Definitely use iNat to learn, and hopefully get good advice on how to ID things you are interested in. If you want more info, put that in your description. Most people will be glad to help you learn how to ID their favorite taxa.
5 Likes
Lots of great advice above, just a couple of points to add as you start on IDing.
-
In my opinion, what is even more important than choosing a broad or a specific ID is staying engaged with your observation. If you put a specific ID/AI ID and then other users (maybe experts) say it is something else, don’t hesitate to withdraw an ID you made in an area where you lack expertise. There’s absolutely no shame in this - it’s part of the learning process. But keeping a disagreeing ID up that you aren’t sure of does slow down the ID process. That said, if you still feel confident in an ID (even when others disagree), you shouldn’t feel obligated to withdraw it - I would suggest starting a conversation with other users about why they are IDing something - that’s how iNat is great for learning.
-
As a corollary, to the above, don’t, however, just agree with others’ IDs just because they seem to be experts or as a way of thanking them. iNat IDs are supposed to represent each user’s independent best guess/knowledge. So leaving your ID as a broad one (like Insects) if that is as far as you can go personally is absolutely what you should do. That said, if another user’s ID inspires you to research a species and learn key traits or other characters that let you make that ID on your own now, by all means you can update your ID.
And one extra thing I find helps that others have mentioned above is that you can note how you arrived at your ID in the notes. For instance, I use the AI sometimes for organisms that I’m totally unfamiliar with and just add a note like “ID based completely on AI” to the observation, so that other users can account for that and know I have no expertise.
8 Likes
I completely agree with this. And yes, do not just agree blindly. Just a while ago, I had an ID I made to Anisoptera (larval stage) identified to species by someone I sort of know. Rather than just agreeing, I asked how that person identifies them to species. I have no clue, so it’s a way for me to learn, and based on what the answer will be, I will agree or not after having done some research.
EDIT: That person got back to me with some great id information. I wrote it down, so learned something, and agreed with the id!
5 Likes
Bearing in mind, one cannot Withdraw an ID on the iOS app. If one uses iOS, the only way to withdraw an ID is on the website. (Strange, but true.)
7 Likes
Lots of good advice here - one thing I can add is to learn the limitations of the AI. For example, it tends to be good at plants, not so good at most insects. (Some exceptions, but they tend to be things like the more distinctive sorts of butterflies which you will quickly learn yourself anyway.) For plants, I will trust the AI if I have 2 or 3 different photos of the same plant and the AI comes up with the same ID for each one. If you only have one photo, even for plants, I’d be wary at trusting the AI to species, at least leave it at genus.
But so long as you stay engaged with your observations and withdraw IDs if someone else corrects them, you can risk listening to the computer vision sometimes, and as other users have said, getting a genus- or family-level ID may well be identified a lot faster than a very broad ID like insecta.
5 Likes
My take on this: Use the name at the lowest level that you can have any confidence in. If you’re not sure of the species and the Computer Vision offers some names, use a genus name if it seems likely. (You don’t have to be certain.) People doing identifications may search on “Plants” or “Animals” but usually at lower levels. If you use the wrong genus (or higher) name, your observation won’t go to Research Grade with just one click and people will be able to find it while searching on genus, family, order, etc.
Watch for your observations on your dashboard and change your identification if appropriate because of other identifications and comments. It will work out.
5 Likes