I totally agree. Inat shouldn’t accept drawings, or at least only as casual evidence. Drawings are not just problematic by themselves but even because open doors to generate images using AI and argue that its an evidence as good as a drawing.
Maybe inat could just open a new section like cientific drawings (simular to a casual mark) that simply doesn’t count as normal data (not entering on IA’s learning data and mapping) but if gets a research level it could be add into species profile like a normal register.
1 Like
As I’ve mention to a few others in private responses, to totally disallow drawings and sketches is to “throw out the baby with the bath water”. Nature drawings and “sketches” should be allowed and judged on their own intrinsic merit. I submit to you this array:
https://www.inaturalist.org/projects/nature-drawing-and-journaling/contributors/gcwarbler
3 Likes
Field guides that only include selected species do so not because they opted for photos instead of sketches, but because of other editorial choices (usually space/cost and/or focusing on the most commonly encountered species).
Photos do not inherently take up more space than drawings, though they may be more costly to print (on the other hand, it is probably often cheaper to find/commission photographs than to commission a skilled illustrator).
I disagree that botanical sketches necessarily show more details than photos, provided that both the photos and sketches are equally well done.
Schematic illustrations are useful for abstracting out particular elements. Sometimes this is valuable, sometimes it is limiting – for example, it is very difficult to convey complex textures in this way. The punctuation or wrinkles on the surface of certain parts of the thorax is often relevant for hymenopteran and coleopteran identification. I find it is much easier to understand what, say, “coarsely punctuated” looks like on a three-dimensional exoskeleton if I can compare this to a photo rather than a drawing. For plants with complex hair structures (say, Pilosella or Verbascum) I also find it helpful to have a photo to compare to.
I have guides that use both sketches and close-ups of relevant bits of anatomy – it isn’t an either/or choice.
4 Likes
Disable drawing is throwing out a lot of great work, that could be used to better explain differences between species. The real problem is only how it is affecting data. I think the same problem applies to museums registers (if im not wrong they don’t count as casual).
1 Like
If people’s primary concern is feeding the algorithm unusable images, then a better solution would be a different data quality assessment algorithm for whether something should be excluded from the algorithm. I’m not sure what exactly the algorithm needs for training, but for example, a photo where the organism is only a small part of the photo can be viable and get research grade (and there are a bunch of reasons we shouldn’t demand people clip things) but might not be good for training the algorithm. But i think the site devs have vetoed this idea for the time being, last i heard.
1 Like
Drawings and description are evidence. Bird records committees, for example, do accept them. I look at them with more skepticism than I do photos, but I wouldn’t necessarily reject an observation that depended on them. (That said, for birds it would be better to submit an observation consisting only of description to eBird, not iNat.) The drawings that are most convincing to me were made while the organism was observed or immediately afterwards – in the field.
What about using AI to make a “drawing” of something you observed? As you guide the AI you change your memories. I’d have the same problem with that that I would have with a drawing made later, after consulting field guides. Plus, the AI will insert details that the observer wouldn’t include in a drawing. I would reject an AI drawing (if I recognized it as such) a lot faster than I would reject a pen and ink field sketch!
Some drawings, like some photos, are not identifiable. If it’s inadequate, revising the ID to “animal” or some other high level with a hard disagreement would be better than starting an argument that basically comes down to the definition of the word “evidence.”
4 Likes
In my experience this is part of why sketches are so risky. Something might not stay long enough for adequate sketching, requiring you to fill in gaps from memory or interpretation. The worst outcome of this is someone looks at other photos or examples of the species to complete the remaining details.
And there’s plenty to be gained for submitting faked drawings, I’ve seen enough of them in the birding community to claim a rarity or add a species to their list (be it a life list or year list).
I’ve seen more misuse of drawings than I have proper use of drawings, and that’s largely why I’m so wary of them being used as evidence.
3 Likes
Thanks, that makes sense. When you mention this misuse of drawings, have you seen this on iNaturalist, or elsewhere? I wonder whether there is less of the competitive listing vibe here compared to some corners of the birding community
2 Likes
I’ve never observed abuse of drawings on iNat myself. Most of the fake observations i’ve seen have been stolen photos from clip art and such. That doesn’t mean it never exists, but making a drawing takes a lot more effort than nabbing clip art from google.
8 Likes
Literally anyone can draw and describe what they saw if they have hands and memories. If they remember everything they saw, I’m sure they can put a drawing into observation with detailed description, then IDers will be able to Identifying them.
So, do generative ai observe nature? I don’t think so. these are just putting data into a mixer and create new false & fake things, not by your brain.
Also anyone shouldn’t using generative ai if they love nature (and you use iNat, so you love nature) these are huge water and energy waste :/
4 Likes
Me too!
How people in past are observed and recorded nature? paper and pencil.
iNat should focus on eliminating fake image & audio created by generative ai and stolen or scraped image from internet, etc.
1 Like
Sorry, not true. Speaking as someone who can’t even make a stick person look right, drawing takes skill - and not everyone has it. Maybe I could develop it over time (though I admit to doubts), but I can’t draw anything worth having.
(It used to frustrate me no end that what was called science devolved into pure art when it came to prac books - I could learn the information, but my best attempts at drawings looked like their ‘shouldn’t look like this’ examples.)
2 Likes
Yeah, not everyone is good at drawing but I think the quality of drawing is not important. I want to say that every drawings are valuable!
Let’s say that you saw a ladybug and couldn’t take a photo.
If you remember a shape of ladybug and spots, you could drawing and describe like this.
I know drawing observations are hard to id than photo observations, but identifiers can id those observations to higher taxon level, at least.
5 Likes
I’m still not convinced that I could draw a recognisable ladybird - but I wouldn’t try. Just to clarify, I’m in no way trying to say that those who can draw shouldn’t be allowed to submit drawings as evidence. I assume that they will be identified to whatever level is appropriate, just like photos - if that means they can never get to RG, that’s fine, but if they can, that’s fine too.
1 Like
That’s totally fine!
I’m also wanted to say that drawing observations take a lot of effort like other photo observations 
Reminds me of teaching Plant Structure years ago. I required the students turn in drawings for the first few labs. I told them, “You could draw when you were six years old! Use those skills. I’m not expecting a masterpiece.” I must say that some of the drawings would fit well among first graders’ drawings on an elementary school wall (except for subject). But that wasn’t the point. To draw an object (a real object), even crudely, you have to look at it, hopefully well.
11 Likes
A good point, and one that I’m not sure I’ve ever really thought about (too bogged down in the impossibility of producing anything good). It still seems unfair for people to be graded on something so far outside their actual field, but for the first time I can see some benefit in forcing students to draw - so thanks! :)
2 Likes
I didn’t grade the students on the quality of the drawing, just whether they drew something. Well, something that with a good imagination I could think might be what they were supposed to see and not, say, a dog.
8 Likes
I agree completely! Identifying organisms based on drawings is a skill, the same as identifying organisms based on verbal descriptions. People make mistakes, yes, and they draw their mistakes, but they tend to do that in predictable ways.
For example, if someone tells me they saw a raptor that was bigger than a Bald Eagle, I don’t assume they saw a Golden Eagle based on that information, because people are notoriously bad at judging size in the field. But I can rule out a sparrow, because people don’t get size that wrong!
Likewise, people often imagine blue and yellow where there isn’t any blue and yellow, because of light and reflections from the environment. However, people rarely totally imagine something being red. If they put red in a drawing, it might be in the wrong place, but there was probably some red somewhere.
4 Likes