Assuming it is right, why wouldn't I use the AI suggestion?

When I upload (Android), the AI suggestions come up quickly and not always but usually include the correct genus if not species. I always click it, although I often downgrade to the level to which I am comfortable (genus, family), so my uploads have that sparkly shield thing that indicates I used AI.

Is there any reason I should be using a different procedure?

2 Likes

Doing that is perfectly fine. I do it all the time as it’s often easier than typing in the name myself.

13 Likes

I see no problem with using the Computer Vision to start off the community ID. I do it all the time, a lot on iNat do. Just be prepared to withdraw it in case it’s incorrect.

5 Likes

Sparkly shields on most of my IDs.
It is That taxon and CV gets me there quicker.

1 Like

Oh, no worries, I am professional level at admitting when I am incorrect, having confused Bee Flies for bees and stingless bees for flies quite often at the onset (more to do with terrible eyes). Now I know on which to s-l-o-w down. I typically downgrade unless I am extremely certain. (I am 99.9% of the time only observing in my own garden so also 99.9% of the time see the exact same species. Consider it like an extremely detailed study of one small plot of land.)

11 Likes

Then you’re all set!

1 Like

There’s another similar thread. I am mystified why people see IDs using CV as lesser. It is a tool like a field guide. It is still our decision to use the CV suggestion, or not.

Rewarding ‘I can typing’ as better value, is, odd ;~))
Perhaps it is - I can spell that Greek and Latin binomial?

9 Likes

The only time I would not use the CV to make an ID on an observation is if someone has already used it to make an ID. I think only one ID per observation should be based on the CV, and that is generally the person who posted it. The other IDers should be identifying primarily their own experience or at least have studied “similar species” carefully. I’m not saying you need to be an expert. just take more than five seconds to study the species.

If you’re just talking about the shield thing, the only possible use of it is to know whether someone typed it in from memory or just clicked it cause it looked right. That doesn’t t necessitate whether they actually are experienced in that species though, just more likely. Usually, their profile pic or profile itself will testify to their ability. I often encourage IDers who have blank profiles to say a sentence or two because it is a stereotype of those who don’t really know what they are doing but just like clicking buttons. If you’re already on the leader board then it definitely doesn’t matter. I’m on the leaderboard on most seashells but I still type them in anyway.

3 Likes

I agree. I never use the CV when making an identification on someone else’s observation unless it is an unknown that doesn’t have an initial ID on it (and then I’ll often note that I’m using the CV suggestion in a comment). The CV is great for getting an initial ballpark ID so that other users with expertise can find the observation. But multiple people using CV on the same observation just leads to a cycle of CV confirmation of itself. If another user hasn’t used the CV (no icon), it may be for good reason - they knew the CV suggestion was incorrect or similar.

I think the CV icon has minimal use for identifiers in evaluating a previous ID. Many users (including myself and plenty of people who have already commented) use it as an efficient workflow even if we can ID from memory without it, so seeing the icon doesn’t necessarily indicate that the CV was part of the decision-making process. I really only use it to evaluate IDs as potential CV errors in three situations:

  1. Seek IDs (where I know that model doesn’t do a great job sometimes)
  2. Situations where I know the CV suggests specific incorrect taxa or doesn’t have taxa included
  3. Out of range observations where CV is still suggesting
4 Likes

When I am pretty confident on the ID (at whatever taxon level), and I start typing and see the name pop up in the CV suggestions, I just click on the CV suggestions.

5 Likes

It’s there so why not use Computer Vision.
It does not determine your species, so if you are not sure it is always better to choose a higer taxon level.
https://forum.inaturalist.org/t/opinion-on-the-computer-vision-icon/40910

1 Like

Why do you value someone who types out
https://www.inaturalist.org/taxa/460167-Leucospermum-hypophyllocarpodendron-hypophyllocarpodendron
I don’t get it. A spelling contest?

4 Likes

Only slightly off-topic: The winning word in the 95th Scripps National Spelling Bee was psammophile, an organism adapted to a sandy environment. Reading the news report I was amused that I had myself learned this word the previous day in the course of identifying iNat observations of plants in the California desert. So maybe the link between accurate spelling and taxonomy is more complex than it appears?

10 Likes

I don’t, sorry, I didn’t mean to imply that. I am dyslexic so naturally am bad at spelling even the common names, but usually, I only have to type out the first few letters before auto-fill. Sometimes I just click the CV’s suggestion, then wait a few seconds and click it again. This removes the shield. It has become more of a habit than an opinion at this point but I certainly don’t think less of people who don’t do it.

3 Likes

As long as the ID is correct, there’s nothing wrong with using CV suggestions.

16 Likes

I rarely have the option of using CV (although occasionally when I’m ill I go on an identifying spree, and by the African species names getting attached to U.S. observations in the subsequent weeks, it seems occasionally the CV uses me), but the only reason I can imagine to get irritated by it is when one person suggests a species level I.D. because the AI suggested it, and someone trying to tidy things to research grade comes along and agrees without actually having an opinion, and as someone already implied, you can end up with a computer patting itself on the back for being right and going down a stranger and stranger rabbit-hole of self-congratulation.

But that can be just as true if someone types out an I.D. that they think is in the ballpark, so it’s hardly a problem specific to CV. I didn’t even realise there was a shiny badge.

4 Likes

This plant taught me that word. When I learn a new species, I like to track down what the name means. Sometimes it helps - sand loving covers most of our area. But not all of it.

3 Likes

Great idea! I didn’t think about that! I bet a lot of names have a whole backstory :)

1 Like

I am glad to be in such esteemed company. I hope I have been clear that I only use the suggestion when one proferred is the species I would have been typing out anyway.

I am looking forward to when the app begins to recognize Eurhinus yucatecus so I no longer have to type that one out – how many of a species have to be observed before it is suggested?

3 Likes

The only reason I can think of is that it might be useful for some researchers to be able to differentiate between IDs provided solely by AI (human clicked the correct ID but didn’t know it was correct) and IDs provided by humans + AI working together (AI provides the name, human knows it’s correct and chooses that option from the list, sparkly badge appears even though isn’t a solely AI ID). I’m not sure it’s worth typing out the name though, especially while in the field.

2 Likes