Human observation shows under dashboard subscription for Psilocybe

In the attached screenshot, iNat is showing me Homo sapiens when I’m subscribed to the genus Psilocybe. Last time I checked, humans are not mushrooms. Is there a reason for this happening or is it just a glitch? I’m using the Safari 13.0.1, if that helps.

Cheers

1 Like

The initial ID was Psilocybe but was IDed as human soon after that: https://www.inaturalist.org/observations/33652850. I’ve had the same thing happen in my dashboard when IDs have been corrected.

3 Likes

Awesome :+1: thank you. I was a little confused when I first saw it

1 Like

I don’t know what it is, I guess I’m not cool enough, but there seems to be some kind of online meme or joke to call people White Button Mushrooms. We get it for humans all the time. Your dashboard will always show the first ID which triggered it.

4 Likes

I don’t think it’s a meme in this case, though, because you can see by the little icon that the AI provided the suggestion for the first ID.

P.S. I could be wrong, but I don’t think the AI is being mistrained by memes; per other forum posts by staff, you need a certain # of RG observations to affect the computer vision, and I don’t think enough of these have been left uncorrected for that to be the case.

1 Like

What icon might this be? :thinking:

1 Like

This little one next to how long ago the ID was made, I kind of love it since it seems to indicate something magical happened. :grin:
06%20PM

4 Likes

kestrel provided the screenshot above.

If you click on it, there is a pop-up dialog saying it came from the AI, like this:

image

2 Likes

I didn’t even notice that :joy: I’ve seen it but never knew what it was. That’s awesome

2 Likes

I am subscribed to guinea pigs and I get a similar thing all the time, people putting pictures of humans and identifying them as guinea pigs.

in the case of the mushrooms there appears to be some odd quirk of the algorithm where it offers that ID as a possible ID when humans are in photos. I don’t really know why, and it’s odd, but i think part of it is that (i think) the algorithm isn’t ‘trained’ on identifying humans since those observations don’t reach research grade.

3 Likes

I think that is right, and unfortunately that tends to promote “joke IDs” among a certain subpopulation of users, by offering them 10 other non-human options to choose from instead when adding an ID to a photo of a human.

2 Likes

iNat has nearly 30 000 observations of Homo sapiens. Training the computer vision to ID Homo sapiens correctly, might help to reduce the number of jokes.

2 Likes

OTOH, humans do seem to sprout up all over the place, spread with no obvious means for having gotten there, and are frequently found where there is decay. There is some debate as to the direction of the causality, but association is definite. And my goodness but there are a lot of them!

4 Likes

The computer vision clearly is trained on humans even if they are not Research grade records as it is almost always the first suggestion. That’s not the issue,the issue is the other 9 suggestions.

I submitted a change request to suppress the other 9 if human was the 1st choice but it was declined

2 Likes

Somehow I suspect teenage boys IDing their teenage boy friends as toucans and mosquitoes (the most recent not-ID IDs I’ve seen) has zero to do with the suggestions available… :) They’re going to find goofy things, whether or not it’s offered. I attempt to drown those new users with super friendly helpfulness and suggestions, using the frequently used responses everywhere possible! Generally I assume it’s just kids not realizing that literally the world is watching.

@kestrel IF ONLY there was a little blinging magic happening sound effect when you mouse over that AI icon!

4 Likes

Honestly, as long as teenagers are using iNat, let them have fun :slightly_smiling_face: while they many not understand the implications of an ID (I’ll admit I don’t completely understand either) at least they’re using the site and are able to enjoy being out in nature. As they mature, they’ll probably learn better. As long as the IDs get fixed, there isn’t too much of an issue. It just surprised me to see a human in my feed :joy: and I didn’t know why that happened

3 Likes

Interesting. A few joke IDs I have looked at lately, of images that are unambiguously human, when I bring up the Computer Vision suggestions for a new ID, human is not listed, but the joke ID that was chosen is the first choice on the list that does come up.

You would be surprised. One recent joke ID I looked at had a clear image of the student’s teacher, identified as a snake species. When I looked at the computer vision suggestions, the snake was indeed first choice on the list, and human was not among the choices.

1 Like

When I use the suggestions and clear the taxon, it always pops human up when I see them. (And judging from at least the “toucan” photo, I’m pretty sure they were chosen rather purposefully! :face_with_raised_eyebrow: ) The ones I’ve run into are usually coming from kids who put legit IDs on their observations after the first couple, presumably they’re posting due to a class project; they know what they’re doing.

I can see the mushroom ID resulting from blindly chosing the first suggestions. Seems like a roundish face on a different background could easily look like a mushroom cap to AI. But a toucan? Or a mosquito? Or a guinea pig (aka test observation subject? :rofl:)? Hm.

2 Likes

Here is another example of a Human labeled as mushroom. https://www.inaturalist.org/observations/33744802