Step 1: Assuming I made that ID consciously and not a spurious ID, i could have seen spider just like the iNat AI did
Step 2: The user rejects that any photos are edited for that observation
Step 3: I have noticed similar cases where some people’s AI ID is very off from images in other observations at some point before, which is making me doubt if this case is worth investigating
Step 4: I thought above is because of taxonomy change/better CV versions but this particular observation link above with very recent ID is causing me doubts of that hypothesis.
I have encountered a similar problem while IDing. I select the ID and immediately move to the next observation, but the new observation loads faster than the new photo, so I still see the old photo for a little while. Doing this a few times in a row, all within a few seconds, can result in seeing a new photo on a new observation, but the photo and observation don’t match. I posted about this a couple of years ago but was told just to go through the observations less rapidly and the problem wouldn’t occur.
I’ve seen this on the Android app (I’m on 1.36.8 (628)), but never the website.
Sometimes if I scroll too quickly while looking at my obs (I’ve almost never used this app to browse obs more generally, so I dunno if it it happens elsewhere), the wrong cover pictures for my obs will be displayed. Scrolling again and doing so more gently usually fixes the problem. Sometimes I have to restart the app.
There’s a second bug that happens if I swipe left or right while editing observations. If I go to edit an observation, and I set a taxon ID from there, and then I swipe left/right, and edit the next observation and set a taxon ID, then when I accept these observations the taxon IDs will not show as their proper IDs. Like, let’s say I picked “Angiospermae / flowering plants”, the result will be that the ID shows “flowering plants” as if I’d just written those words manually as a custom placeholder ID rather than chosen an established taxon. My workaround is to edit only one observation at a time, i.e. accepting the changes to one before going to another observation. Anyway I don’t know if this second bug is related to the first one, but just posting it here in case it is.