AI disclaimer & Free lesson: How to spot AI
Because someone asked: I do NOT use AI for my writing. Unless you mean me handing it my fully done outlines (they're way too big) & ask it to cut it down into something manageable for easier overview.
A(n extensive) note to the Cover Art for Book 1 (or rather my placeholder, as I hope I will get an actually handdrawn one done some time later) since someone asked:
In the spirit of full disclosure:
The current placeholder Cover Art for book 1 is actually an AI-image.
Albeit one prompted by my own art piece (drawing / painting)…
…plus a lot of finagling with the text prompt on top of my drawing via countless (literally hundreds) of random attempts with style, version, agent, random generation until finally one just went: O.M.G., this one actually works.
.
Yes. The horror. If that makes you leave—you’re free to do so.
But no, I do not use AI for writing just because I used it for a single picture I could not get done any other way, the way I wanted it to (yet), as some are so prone to suggest.
In fact, I don’t normally use it for pictures, either.
.
There is no real follow-through logic to this assumption.
Take it from an autistic person.
There is no rule that says if A then B as well.
Sure. You can argue that most people who are willing to use AI once, will be willing to use AI more than once — but the specifics of HOW and WHY they use it are not as transferable as easily. Not by far.
The CATEGORIES are unlikely to just randomly shift on what you are willing to accept and what not. Photography/Painting vs. Writing are not even adjacent things, you know?Unless you mean in a VERY broad sense.
And the broadest sense is… well. Basically unusable to draw from for such arguments. Overlap usually happens in very close things, not in things very far from each other, if we’re talking about general likelihood.
.
Yes, I’m probably a horrible person, you’re right. I’m also just a struggling human.
.
That one had become a resource problem down the line after over a year of trying otherwise (see below). Actually, several years.
.
.
The AI made this one WAY better than I can yet do when it comes to people (like: you would probably laugh at the original; I am not good at drawing people — the dragon and such? those are my own, painted with real physical paint, in case of the dragon that’s actually wall art done with normal DIY market paints, two colors (blue and green) mixed with varying degrees of each other and white to get a variety of shades). This is undebatable. It succeeded (among hundreds of fails) to actually produce something that FINALLY let my autistic brain rest in a satisfied: YES, that is him.
I still want to do my own at some point, but I am simply not there yet.
.
.
Before you scream about “get a professional”:
I have ZERO income atm. Literally zero.
I have been in neurodivergent burnout with multiple chronic illnesses for over a year now and am currently in a legal battle with the state about disability and unpaid welfare cheques for existential help in a crisis.
I have 6k MINUS on the bank, overall. It would be more if friends hadn’t pitched in. It is that bad.
.
Yes, I have art friends.
But: No one even tried to make one — they don’t have the time.
I did not have any other offers.
.
I don’t have the resources (on many levels) to get one from the internet, trying and trying again.
In general, I also have a massive psychological barrier against begging people for free stuff. (I suffer from c-PTS syndrome as well as neurodivergent RSD.)
This ten times as much on the scales when I might then reject what they give to me afterwards because my brain goes again “sorry, this is not it” — which uses to massively disappoint or even anger people, when I then end up not using it. And no, I will NOT use a piece of art that does not fit what my brain is screaming for. It takes away too much from my capacity as an incessant annoyance in the back of my head, think: an open tab with annoying spam music loudly blaring from it that you can’t get rid of, taking away processing power from your brain. Mine already has less than those of other people (or rather: it is much more occupied with stuff that other people have barriers against — mine are non-existent).
And no, I cannot ‘just pay them a fair price’ as I would dearly love to.
While I have internet access via a friend, I have no money left for non-necessities.
Next month, I might not have enough money to pay my rent anymore. To pay my food or my meds anymore.
.
I did, however, have access to AI via my art studies still (that I will likely fail due to my physical collapse now).
.
And I DID search for free material, too.
(I’d even searched for artists before, when it had not been a time-pressing issue yet, but couldn’t find any I liked well enough for what I wanted. It happens. Yes, maybe I didn’t force it enough. I had no pressure to do so back then. I hadn’t even written the full first draft yet, after all. This was just an RPG character I liked and wanted a picture of. Of course the investment wasn’t as high. By the point the pressure ratched up a few notches, my resources already started dwindling. It was still not a primary concern.)
Being AuDHD, when I say I searched, what I mean is: I searched for YEARS, a total of at the very least whole days of actual hours summing up to full days, possibly weeks.
I searched even more forcefully once the first few drafts were done.
.
I could not find anything that screamed “this, this is him” like this one eventually did.
Not even close. In fact, not a single one that got even close to my mental picture. Nada. Within YEARS.
.
Then, somehow… This one popped up. I fell in love. Yes, I fall in love with an AI picture since that one among countless slop actually got the “THIS” reaction. I had no control over that. I still don’t. When you know you just know. This one actually fits my mental picture of him. I’m sorry. I didn’t choose this. In fact, I stumbled into this.
.
So how did I get this one?
I goofed around with AI at some point during my art studies, just to try it out.
Since I had no real idea what to go for to see what it can do, after random landscapes and what-not, I eventually attempted one for a beloved character, too.
One I had already been unable to bring to the page on my own despite having a very clear mental image that makes my autistic mind reject any that do not “fit” this image, don’t even come close.
(What I usually do are landscapes, maybe some animals.
So my own art is just not there yet when it comes to people, and I cannot get the image out of my head and across to someone else in that detail. I know exactly what to change but my hands are unable to do it on their own yet. It is frustrating to no end.)
Then this happened.
During hundreds of random attempts, seeing if the thing could get there (thinking it very likely won’t), what it does when other styles are used, what changes with the text prompt changing next to it etc.
Why? My school actually even asked us to try it out, so we learn to spot what it does and how it works (and because it is not entirely opposed to it, but says it’s ok to use as a tool for help to some degree; why not? the world is clearly going there no matter what—while one can debate that endlessly, I will not do so here, but in the spirit of full disclosure, that’s the background to this picture).
I got this one nearly by accident, when I had almost given up on the idea that even AI could ever do it.
But it did.
I feel conflicted about this myself. I have very much not been a fan of AI. I know the legal troubles, the way one can never tell anymore which works went into it that might have been scraped illegally and without reimbursement. I know about medical files that somehow ended up unprotected and got real people put into a database they should never have ended up in. Worrying stuff.
I was actually openly against any AI myself before I got pushed into the “doodle some” side task to one of my projects (didn’t use anything AI in my projects but a single fake-label in the fake-l’Art Nouveau-style, and yes, we always label where we used something not our own and what is AI (have to; of course we have to).
After finding AI a helper against melt- and shutdowns in crisis where I don’t have anyone able to help nearby, I can see it has some actually good use. I still don’t know what to think of the picture help. I guess I will just remain eternally conflicted about it unless AI companies reimburse the real people they scraped from, large style. But even my school taught it as something becoming basically unavoidable and better to know how to handle, better to know how to recognize and use with rules, aka: within limits. As a helper, not the artist. But yes, I know it is still replacing people. Especially in the departments I had been studying for, to make the joke fuller and more horrid.
And here I am — as of now unable to avoid it myself in my free time unpaid project because I just love how the character turned out. I want my own version. I am unlikely to get it any time soon.
.
If you’re reading this as an artist willing to try your hand for free?
I’d be eternally grateful for even the attempt.
.
But please be aware that I am unable to reimburse you in my current situation (unless as an art exchange, if you need writing or anything else I can do).
.
I’m fighting for my right to exist right now. I don’t have the capacity or resources.
Worse: My autistic brain might even end up rejecting what you might make as “this is not him”. And I can’t do anything about that. It is not a choice, least of all a conscious one. It either fits or it doesn’t.
—> Please only consider this, if you can live with me potentially rejecting the piece
without needing to vent your feelings with me.
I am in autistic burnout. I am unable not handle that.
.
And my brain is… very specific about this one.
I know exactly what this character look-feels like and if it isn’t it, it just isn’t it.
This has nothing at all to do with your skill or my appreciation of it.
My brain already rejected THOUSANDS of pictures and photos of & by real people, real artists AND AI before finally lighting up on this one. So you’d have to get REALLY close with the character and capture him just right or my brain will not accept the replacement.
It’s the same way I am unable to stand tags in clothing and cannot ignore that smallest electrical buzzing you likely wouldn’t even ever hear.
I repeat: This is NOT A CHOICE.
.
.
.
Free lesson—How to spot AI:
In this picture, you can see it the easiest from the blank space where none should be, the way this is set up (people don’t usually leave this little as entirely blank if they do leave blanks for some artistic purpose; I just didn’t have the time and energy yet to work on art again to try and remedy some by hand again from this latest attempt) and how the style is not entirely consistent, e.g. parts of it are too crisp compared to others.
.
Also: Lack of transitions (see below). The image has the same problems many photoshopped pictures have: If you cut out one thing from one picture and put it into another, the lighting etc. is not quite right, and it still looks as if pasted onto there.
.
If you look really carefully, you can also see it in the eye being still the slightest bit strange, more than just because it is a very light color (as it should be; they guy has stunningly light gray eyes), at least if you zoom in a lot—though this is one is a near-perfect version when it comes to an eye.
Which is very rare for AI, btw (especially in AI images where you can see the whole face, not one turned aside).
AI has a propensity to blur eyes. Make some kind of molten version of eyes where two eyes kinda meld into each other or the eye sits strangely in the face or the pupil strangely in the eye and such.
.
AI will also give you too many fingers or random pieces not attached to anything or cut off (the most easy mistakes to spot), the height / width of people will be off, there may be spots where you can very directly see where two pieces of picture were melded et cetera. Rather often, you can also see what I said for the above already: Style mixes that remain visible within the same picture.
Here’s an example where I used the AI help to try an insert a different bird. Can you see the strange accessoire it put on his head while doing so because it got confused about how to transition between the parts here? (Hair / Lower left wing of the big right-side bird): It’s got the wrong angle, the wrong size, the wrong style—in fact, I am rather sure, this is a belt buckle inserted into the pic that the AI turned into a hair clip in its panic to try and fudge the overlap of the two parts somehow.
.
Also pay attention to how the crispness is changing without transitions in between and for no reason (foreground/background, intentional lenses to direct the eye to more important spots etc.). The photoshop problem I mentioned. Some of the birds are too clear. The character especially is way too photo-like in comparison to the background technique. A real artist, in the very same picture, is likely to make a more coherent look, unless in rare outliers (working with very long phases in between, intentional style mixes for a certain effect in collages).
.
A word of caution, though, before you think every not-quite-consistent style always means AI:
Kitbashing often has similar problems, unless the artist is already VERY skilled at removing any trace of the mix that was done.
(And some artists may intentionally use different styles within the same collage, too. Though that… looks somewhat different. I’m having trouble describing that, though. Sorry. I’m just a student and hobbyist. There are probably people who could describe this to you in detail. I don’t know any, this is just my pattern spotting going “I can see it”—I cannot name it, however. Or do it, for that matter. This is just a random thought right now, I don’t have examples at hand. I’m sure you can find some via search engines if you truly want to.)
.
What am I talking of there and how is that different?
Kitbashing (or photobashing) is what the process is called when artists use (often freely available) art (or one they’re otherwise allowed to use by other people, though this can also be done entirely with one’s OWN art and would still be called the same) to assemble their own collage out of snippets and pieces from other art, painting over it with e.g. Photoshop to get a unified look, adding their own pieces.
And yes, this is a common technique that artists don’t usually frown upon. (Though some might.)
.
Matte painting for movies does a similar thing and being able to do it well, well enough people don’t notice, is actually a real skill—good matte painters will have you believing you’re seeing a real place while in truth you’re seeing something handpainted over parts of actual shoot! It’s crazy, the good kind of crazy.
AI is doing something similar here, though not quite as well as a real high-level professional could.
.
.
And of course, people not being at the necessary skill level yet will ALSO have structural problems of sizes not quite fitting each other (some body parts being too large for the whole etc.), that’s a normal student problem.
.
Melding two eyes into one attempted one or adding too many fingers or cut-off random pieces with no or little attachment to anything else usually isn’t though. =P
—> If you find something like that, you can be mostly sure that piece was at least made with the help of AI. ;)
(Photoshop, for example, has an included AI help now (you don’t have to activate it, it’s its own function). The “magic wand” has always been a bit of a similar thing. But the new AI actively generates stuff not just tries to make transitions between two parts of a picture or fill up something with what you’d like to have replaced with another part from the picture.)
.
.
When it comes to text, however:
You’re basically fat out of luck unless someone has such a unique style, it has not made its way into the AI database yet.
Em dashes are NOT a sign of AI—a lot of authors use them and have been using since before AI ever existed.
***
Weird placement? Well… AI didn’t only learn from “the greats” but from anyone it had free access to on the internet.
Which means it also learned from dyslexic people and people who don’t conform to the old grammar rules.
And those still exist.
Of course dyslexic people and also people who grew up with simply a DIFFERENT formatting / grammar style that you might not know because they live in / come from another country… well? Yes, 100/100: They still exist. Like, everywhere.
It’s not like AI has purged out any differences in humanity yet.
Just because YOU learned one style… doesn’t mean someone else won’t have another. Just because YOU have come past a certain point in your writing… doesn’t mean everyone else has.
Just because YOU… well, you get it. Right?
AI is not the only one who can write texts that you perceive as weird—or actually sometimes really good ones that you might not perceive as such at all. It’s all a bit a matter of luck in the generation and what else you’ve encountered.
.
The one way you DO know if it was written with the help of AI or not?
YOU were the one writing.
AI mistakes are easy to see when you chat with it.
You can spot where it parrots your own phrases back at you, mirroring too much in its attempt to make you comfortable. Or where it forgot stuff from X chat posts before, mixed up names or whatever. (Hopefully.)
But it’s not like real people don’t have pet peeves. Or as if real writers don’t sometimes mix up like… basically anything from their own worlds and writing, depending on how long they haven’t interacted with it or if it was something minor (which is easier to forget, especially if you have a massive world and sprawling cast, let me tell you).
Ever met a GM/DM at an RPG group who forget how their NPC was named? There.
That’s what I’m talkin’ about.
If you feed AI enough of a specific voice, I bet it can even start to parrot back in that very unique voice. It’s what it does. You’ll miss a lot.
Meanwhile you ALSO miss a lot of real work and insinuate people who never touched a single chat tab of GPT or Claude or whatever its name as “fake”.
On no real basis. Because:
Even the best fact checkers and pattern seekers can’t tell.
.
BTW, did you know that grammar rules are all just willy-nilly choices with an army behind them? You can thank my linguistics professor for that one.
But it’s true. Historically, it was just some old white men who decided “this is how we’re gonna put it into books — and then we’re gonna make everyone have to read and use them or we’ll call them weird, non-conforming, and generally ridicule and ostracize them”.
This also included the so-called “generic” he…
…which is absolutely NOT “generic” at all.
***
It was a choice made by old white men of a certain elite circles at a certain time (long gone by now) and they were FULLY aware that women (or anyone else) would NOT feel included by the prescription of the word “he”.
***
They openly talked about it in their private letters, which is how we know.
Although patriarchy was so good in hiding that fact and spreading falseness that nowadays a large part of English-speaking populations STILL think that “he” is appropriate or that “he/she” have to be used in variation, ignoring the ABSOLUTELY CORRECT, ACTUALLY GENERIC word that had been around for CENTURIES before that conscious attempt to try and remove it by installing “he” instead. I am speaking of the generic SINGULAR “THEY”.
Singular THEY is the correct generic term, folks. Use it.
.
We came far from AI now, did we? Maybe not. ;)
As I said, grammar (and vocabulary) rules are all basically random choices.
In older times, people wrote however the fuck they wanted (if they could write), according to what they thought sounded most like the word they knew from speech production. And those words change a lot over time, with cultural contact, influx of immigrants, trade, the sheer LAZINESS of people etc. etc. — there’s actually a lot of research on HOW languages change (they basically go from exploding with categories and extra-additions to be super-SPECIFIC back to trying to be as SHORT as possible, which loses specificity again, and thus the circle repeats) — how does this relate to AI and LLMs (large language models)?
.
Well, turns out those tools people try to use to find out if something is AI or not …
(which are, ironically, ALSO …. drum roll, please… whaddaya think, hm?
YES, they are also AI themselves!
basically AI is told to watch if this or that was done by AI…
AND on top of that… this might very well mean that people put YOUR work INTO AI in the first place, to try to CHECK for AI, even as it might otherwise never REACH AI if you kept it far away from it — new kinds of horror, eh?)
… and those tools? Even THEY cannot accurately say if something is AI or not.
.
Need proof of this?
Google for what happened to Mary Shelley’s Frankenstein and the US constitution when they were presented to AI.
What label do you think they got?
100% natural, because they were written WAY before AI existed?
AHAHAHA—No.
AI-checkers labeled them as “almost certainly AI”.
90-98% (or in some cases even a proclaimed 100%, if I don’t remember incorrectly) probability of being AI and such things on classics and even old legalize…
Imagine that.
So, no. We sadly have lost any way of telling if a human or a machine is writing that could be actually “proven” other than our feeling. And I for one don’t wanna know just how often that feeling is very, very far from the truth.
.
If you look at the picture above… you can see we are already losing control when it comes to telling which images were created by AI.
Though, in some ways… AI might save us from that. Instead of getting better, if it doesn’t get new input? It will start gaining artifacts instead of eliminating them.
.
The most worrying category is not art, anyway, despite how it extracts and often enough outright steals from people (in a world that is, let’s face it, still extractionist by default, via the choice of people hoarding resources and outright being malignant).
It’s your photos.
It’s your faces and bodies landing on the net, because some doc forgot to put airtight security on it and UPLOADED it onto the web somewhere. Or because someone malignant put you up there. And the most normal case: because YOU put them there.
And then someone else, someone malignant, a HUMAN, asks a machine to make porn of you. Or maybe a sandwich to laugh about. Or a sand witch if you’re lucky.
.
Then there’s the HUMANS who ask the machines to make FAKE videos... for ads. With fake claims.
Or worse, because of the sheer scale of impact (though the personal cases are very much horrible in quality, too—this one, though, is about sheer quantity of potential effects): Faking news reels.
E.g.: Fascism(-related movements) feeding you propaganda about immigrants taking your jobs, the “war crimes” of the newest nation they want to bomb that might not even be real anymore… Though there’s a case to be made that they don’t need AI for that. All they need to do is hide reality from you by bombarding you with an overload of trivia and propaganda drool (which doesn’t even need pictures, apparently, because hate and -isms still seem to work well enough) wanting you to forget all about the concentration camps where they rape people, even the little children, and kill people (yes, again: the children, too, especially the ones who got pregnant too early and are now “legally” kept from an abortion that would SAVE THEIR LIVES while the birth is very, very likely to kill them, without anything else surviving either but the LIE, their death burying the truth), maybe only by horrible neglect, maybe outright “ending”.
That’s why good old journalism is more important than ever — fact-checking, fact-checking, fact-checking—and counteracting the propaganda.
I would not, however, trust Fox News or yellow press to do that… would you?
In the end, the last recourse there may be to go there and see with your own eyes.
.
For other things and general restriction of how AI can be used: A lot of legal action.
(And we all know how hard it is to ever get anything deleted from the web again, no matter how harsh the laws and how well your prosecution goes.
This is how lives get destroyed.
But it’s not quite the tool that does it. It’s the PERSON doing it.)
.
Right now? I don’t think anymore that AI art is the biggest problem here.
Art is at least entirely fictional…
.
The war and human rights abuse profiteering is not.
Be mindful what you use your energy for—for me, this here… this is recuperation, to a degree. I wouldn’t even write this if it weren’t. If it took instead of giving something.
But our rage is better used elsewhere than impotently in some web forum, unless it’s organizing and uniting there.
Our RAGE needs to go into the streets.





