Description of problem:
I don’t know what’s changed this year, since this used to work just fine last year. But any observation that I’ve added this year has had completely messed up suggestions when submitting them. Most commonly suggesting a bat for anything - if it can find any suggestions at all. Bird? Fly? That’s a bat, or possibly a vole. Tree? Also a bat.
I’m located in Western Finland, if that helps.
Most importantly, suggestions work fine once I go through My Observations → Suggest an Identification. It’s just Upload → Add Observation that’s failing.
Here’s the suggestions for the same photo from My Observations:
I sent a Help request thru. @carrieseltzer is looking at it.
We had African bats - no - not a bat in sight. But unwanted extra work for our bat specialists.
Birds at Strelitzia - no - not a plant, crane flower is not a bird. This is worse because it adds Kingdom Disagreement and pushes the ID to Life.
I was wondering if the new CV suggestions were just me - but yes, excluding any photos I’ve uploaded that were of…a certain dodgy quality, I had also noticed a rise in bonkers suggestions for things that, to me, looked relatively straight forward.
So it’s not just me? I’ve been watching the bug forum, waiting for someone else to report this, but figured that something must be wrong on my end since nobody did.
We had inaccurate suggestion on iPhone app yesterday. We use CV sometimes so we don’t have to type the species name. Knew it was a Dark-eyed Junco but suggestions weren’t confident or accurate. Toggling off “show nearby” then gave correct suggestion
I doubt that’s related. No matter what’s expected nearby, it shouldn’t confuse a tree or an obvious bird with a bat. And this has been a long-running issue for me, at least since February.
For the people that are experiencing this issue, could you please share what operating system and version you are using (Windows 10, Windows 11, MacOS 14.4, etc), and what browser and version you are using (Chrome 123.0.6312.107, Edge 123.0.2420.53, etc)? Browsers usually have an “About” page where you can see the version. Having this additional information will help us narrow down the cause. For example I cannot recreate this problem on my computer, and I suspect there are some differences introduced by different software involved. Thanks!
Windows 10 Pro, using LibreWolf 122.0-2 and Brave 1.62.165 (Chromium 121.0.6167.184 - ).
Brave’s pretty old since I switched out of that a few weeks ago and haven’t had a reason to update since.
Both screenshots for those who said it wasn’t working well have their settings set to include options that aren’t nearby. Try clicking the “Only view nearby suggestions” at the bottom and see if that cleans it up a bit. There might still be some weird suggestions but at least it’s not looking at options from all over the world.
Plenty of observations of Covid, mostly of antibody/antigen tests. So if a bat looks like an antigen test, it will get suggested. Either that or the AI is achieving sentience and confirming that the pandemic was definitely a crossover event and not a humanmade virus leaked from a lab.
This has happened to me yesterday. Last night I uploaded 24 observations from my moth-walk and I think in 3 of them it suggested something wildly off. Twice I just typed in the correct name, in one moth I knew I had seen it before, but couldn’t remember the name. A bit annoying when the CV suggested a snail. Today everything worked as normal.
Mac Book Air with OSMonterey, Safari Version 17.4.
Sometimes. And sometimes not. Doesn’t make any difference, all suggestions are 100% incorrect in the upload interface. When I try the same photo again, it usually generates a different random list. (Though mostly bats.)
Case in point: a goofy photo of a blackbird that got too close to a camera.
There was ~5 seconds between these checks.
(And like I said, the photo quality isn’t the problem. Everything gets identified fine if I check the suggestions via the Your Observations page, after the upload.)
