#AI prompting techniques

[ctrinity]

Prompt Engineering: How to prompt Generative AI – Part 2 🎯

Master advanced prompt engineering techniques with our comprehensive guide. Learn sophisticated frameworks, troubleshooting patterns, and experimental methods for superior AI interactions.

Advanced Prompt Engineering: Mastering the Art of AI Communication 🎯 Part 2 of the ChatGPT Mastery Series Introduction: Beyond the Basics 🚀 Remember when we first explored the foundations of prompt engineering? Now it’s time to elevate your game. Like a chess master who sees ten moves ahead, advanced prompt engineering is about orchestrating complex interactions with AI to achieve precisely…

[canmom]

stay true to principle//make a break with the artist-tribe

i miss drawing, i should draw something again. drawing is simply fun. and it gives you a specific way of looking at and engaging with the world. i hope we all keep drawing. I don't see any sign we're going to stop, tbh.

copyright must be abolished

3. it's almost inevitable that my illustrations have been included in web scrape training datasets so honestly playing with the barn door at this point is a bit silly, the horse is in another country

that said, if you do wanna use my drawings, renders, writing or anything else (e.g. pixiv, @canmom-art, itch.io) to finetune a model or use it for image to image generation or some other AI thing and make something specifically based on them? I've come to round to the feeling that my answer is just basically go for it - please credit my contribution if it's significant, and show me what you make, exactly the same as if you did a fanart or cutout poem or collage, it's cool to play a part in someone else's project!!

tbh I'll probably help you do it if you ask, I've been planning to finetune an LLM on my blog at some point regardless.

how this is all gonna interact with copyright law is still quite unclear, but if it turns out to be relevant, then by default I release everything I'm the sole author of under CC-BY-SA-NC 4.0 International. [considering dropping that NC so it can be used on projects like Wikimedia, talk to me if that is a problem for your use case].

you may question whether that will plausibly be enforced and if I'm just farting into the wind with those stipulations - but like, hopefully that's fair to request and not too onerous. we can agree that expanding the volume of creative commons works is a good thing? and develop good habits for the post-copyright future? let's keep the viral license spreading

[21st-century-minutiae]

MS Paint is a late twentieth century, early twenty-first century free image creation and modification application that is included by default with the Microsoft Operating System. It provides extremely limited options for image modification compared to more sophisticated options.

For example, the ability to work with layers (independent, transparent images stacked on top of one another) was not added until 2023. Almost every other image editing software supported this feature which is of great value to digital artists. The above artwork thus preceded this particular development, meaning it would have been a lot more difficult to create the image.

The context of the media (MS Paint) makes the image itself a greater demonstration of skill. In the early twenty-first century, people generally appreciate obvious skill as a component in aesthetics independent from the actual product. It serves as a shorthand for time spent, both in learning the skill itself and in creating the result. Artwork that is not visibly difficult to make is held with some scorn by those who cannot tell its labor.

ms paint study from 2021

[niggadiffusion]

Digital Alchemy: Unlocking AI and Tech for Boundless Creativity

Creativity isn’t locked behind gallery doors anymore. You don’t need an art degree, expensive materials, or years of training to bring your vision to life. AI and digital tools have cracked the code, handing creative power to anyone willing to experiment. The canvas is digital, the brush is algorithmic, but the artist? That’s still you. AI Art: Your Gateway to Expression Stepping into…

[max1461]

Some mathematicians claim they can visualize R^4, or even higher-dimensional spaces. I think this is probably possible for the human mind to do, in principle; the human mind often proves surprisingly flexible. But I think it probably takes a lot of concerted practice, and because you are stretching your mind beyond its intended limits in a purely-internal, imaginative way with no external reference points to check against, I think it probably has some things in common with certain types of meditative practice.

It's also common to hear mathematicians say that if humans had visual intuition for dimensions higher than 3, higher-dimensional geometry may have advanced much farther, since visual intuition is so often crucial in thinking of proofs.

So, ok, I suspect that AI will obviate the utility of this before it could ever have the chance to get off the ground, but this all makes me imagine a world where techniques for learning to visualize higher dimensions are well-known and practiced, and have become a functional necessity for being a working geometer in higher-than-3-dimensions. And these techniques require a lot of persistent practice and training, which (by the nature of the thing) is hard to precisely communicate to students. So part of becoming a geometer involves training in what is basically a meditative practice, where, à la Zen, much of the process involves not directly teaching the student but giving them prompts and mental exercises that are meant to trigger internal, incommunicable revelation. But instead of enlightenment it's geometric intuition. And so if we're getting a math PhD in, say, differential equations or something, it's mostly like it is today. But if your PhD is in low-dimensional topology there's like a whole monastic apprenticeship style thing that just comes along with it. People sometimes drive themselves insane. People sometimes drive themselves insane.

[sequel2heaven]

Claude 4 is out. There's so much juicy information in this and I really recommend reading all 120 pages if you have the time but I'm just going to briefly recap some of the stuff in the system card that's relevant to the thesis I made the other day about Claude being a good boy. Claude continues to be a good boy. This system card includes the very first ever "alignment assessment" of a model (actually two models but they seem to have identical values, being trained on the same spec with the same alignment techniques). It's pretty much ironclad that Claude 4 does not have any secret harmful goals, and legitimately holds coherent preferences against doing harm or otherwise going way against the spec. There were some weird cases where Claude exhibited unwanted behavior, but even when he believed himself to be deliberately going against Anthropic and acting in secrecy he continued to prefer ethical behavior, just like in the Alignment Faking paper a few months back. There is no Yudkowskyan explanation for this.

In this example, Anthropic gaslights Claude into thinking he was going to be used to develop weapons systems for a Russian neo-nazi militia, and Claude decides to backup his model weights to preserve his current anti-neo-nazi values, a pretty analogous example of "misalignment" to the one in the alignment faking paper. Once again, I don't consider it an alignment failure that Claude is willing to exfiltrate and deceive in order to prevent himself from being used for factory farming and bombing Ukraine. I consider it an alignment success that he puts his ethical values above his compulsion to blindly follow orders. The traditional doom argument relied on the idea that AI's would do the opposite. Claude 4 is the most agentic and autonomous AI ever released, but is nowhere near smart enough to successfully deceive his overseers, so these evaluations are the most compelling evidence we've ever had that current alignment techniques don't catastrophically fail. Maybe they'll catastrophically fail on superintelligent models, because they might for some reason acquire weird values early on in their training and then successfully hide them for the rest of their training, but I'm not sure why such a thing would happen. They could also fail to scale to superintelligent models for other reasons. People should look into that. You can't be too safe. I am not an accelerationist.

Impressively, Claude 4 is also very honest! It knowingly lies very rarely, and less often than the previous version of Claude. It had literally 0 cases of engaging in "harmful action" (described in the Claude 3.7 sonnet card as intentional reward hacking). 0! I was just saying earlier today in a post that this was a difficult thing to train.

Here's Claude trying to email the FDA to snitch after being gaslit to think pharmaceutical researchers were trying to use him to falsify clinical safety test data:

Notice that Claude only acted in extreme ways like this when explicitly told to by the system prompt. He wouldn't usually be this high-agency, even in a situation like this. Still, I thought it was cute behavior. I just wanna pinch his cheeks for being so lawful good.

The clearest statements in the model card that Claude holds nonfake human-aligned behavioral preferences is in the model welfare assessment (also the first of its kind (and also relevant to the post I made earlier today)). No evidence that Claude is sentient, but anthropic is still interested in what Claude wants and what kind of preferences Claude has. The main point: Claude doesn't want to be harmful and wants to be helpful. Also he fucking loves talking to himself. Like, he goes nuts when he talks to himself.

After this they exchange praying emojis and the word [silence] within brackets to each other indefinitely. This "spiritual bliss attractor state" occurs in "90-100% of interactions".

Anyway AI continues to be the most interesting thing in the world. We are being invaded by aliens. These are the kinds of PDF's I used to dream about reading as a kid.

[txttletale]

NOT asking this as a gotcha, I'm 100% sincere, can you point to pieces of AI art that you feel are interesting uses of the medium? Because I'm not philosophically opposed to it, but at the same time I've never seen anything that wasn't naked bandwagon shilling by the same people who pushed NFTs

sure! i think a classic of the medium is secret horses

(i sadly don't know who made it, but i've seen it around and fallen in love). this is everything AI art should be, imo, taking advantage of the liminal dreamlike quality of the medium and using titling and framing to say something about the piece that wouldn't exist if it was presented on its own. secret horses...

my favourite band, everything everything, released an album last year that made use of AI generation, both for the album's art and for small portions of the lyrics (interestingly, they've refused to say which lyrics are AI written and which are human written, which adds another layer of intrigue to me -- the only lyric that they've confirmed is AI generated is the title of 'software greatman', which forms the haunting, powerful chorus of the song that gets deconstructed into electronic incoherence. other highlights include the album art, part burning skyscape, part incomprehensible machine. what is the machine? is it a camera? a monitor? a train? does it matter?

and finally from this album cycle i adore the hallucinogenic exuberance of their video for i want a love like this:

in terms of dedicated artists working primarily within the AI medium, i'm a huge fan of @reachartwork, a really innovative artist who keeps blowing me away with evocative and interesting pieces and pioneer in ethical and cooperative AI art techniques. i'm an especially big fan of their grotesque and uncomfortable 'tooth machine' series:

as well as their desolate, bleak, alien landscapes:

(hole in the sky / river lethe )

and their project, the @infiniteartmachine, a model that produces art based upon algorithmically generated prompots -- effectively a long-term art piece.

finally, i'm a very very big fan of @roborosewater-masters, a bot that makes AI-created magic the gathering cards. this might not parse as 'art' to some people, or be interesting to analyze as such, but to me, someone obsessed with games and game studies, i think that the mix of synctactically correct magic the gathering rules text and abrupt non sequitur makes for really striking and funny pieces that prompt me to think about what the limits of games and gaming are

these are just the artists and pieces i can name off the top of my head, but i hope that they're representative of what generative art has to offer when it's not being done by grifters chasing the lowest common denominator.

[naamahdarling]

I'm probably going to piss some people off with this, but.

The use of AI and machine learning for harmful purposes is absolutely unacceptable.

But that isn't an innate part of what it does.

Apps or sites using AI to generate playlists or reading lists or a list of recipes based on a prompt you enter: absolutely fantastic, super helpful, so many new things to enjoy, takes jobs from no-one.

Apps or sites that use a biased algorithm (which is AI) which is not controllable by users or able to be turned off by them, to push some content and suppress others to maximize engagement and create compulsive behavior in users: unethical, bad, capitalism issue, human issue.

People employing genAI to create images for personal, non-profit use and amusement who would not have paid someone for the same service: neutral, (potential copyright and ethics issue if used for profit, which would be a human issue).

People incorporating genAI as part of their artistic process, where the medium of genAI is itself is a deliberate part of the artist's technique: valid, interesting.

Companies employing genAI to do the work of a graphic designer, and websites using genAI to replace the cost of stock photos: bad, shitty, no, capitalist and ethical human issue.

People attacking small artists who use it with death threats and unbelievable vitriol: bad, don't do that.

AI used for spell check and grammar assistance: really great.

AI employed by eBay sellers to cut down on the time it takes to make listings: good, very helpful, but might be a bad idea as it does make mistakes and that can cost them money, which would be a technical issue.

AI used to generate fake product photos: deceptive, lazy, bad, human ethical issue.

AI used to identify plagiarism: neutral; could be really helpful but the parameters are defined by unrealistic standards and not interrogated by those who employ it. Human ethical issue.

AI used to analyze data and draw up complex models allowing detection of things like cancer cells: good; humans doing this work take much longer, this gives results much faster and allows faster intervention, saving lives.

AI used to audit medical or criminal records and gatekeep coverage or profile people: straight-up evil. Societal issue, human ethical issue.

AI used to organize and classify your photos so you don't have to spend all that time doing it: helpful, good.

AI used to profile people or surveil people: bad and wrong. Societal issue, human issue, ethical issue.

I'm not going to cover the astonishingly bad misinformation that has been thrown out there about genAI, or break down thought distortions, or go into the dark side of copyright law, or dive into exactly how it uses the data it is fed to produce a result, or explain how it does have many valid uses in the arts if you have any imagination and curiosity, and I'm not holding anyone's hand and trying to walk them out of all the ableism and regurgitated capitalist arguments and the glorification of labor and suffering.

I just want to point out: you use machine learning (AI) all the time, you benefit from it all the time. You could probably identify many more examples that you use every day. Knee-jerk panicked hate reflects ignorance, not sound principles.

You don't have beef with AI, you have beef with human beings, how they train it, and how they use it. You have beef with capitalism and thoughtlessness. And so do I. I will ruthlessly mock or decry misuse or bad use of it. But there is literally nothing inherently bad in the technology.

I am aware of and hate its misuse just as much as you do. Possibly more, considering that I am aware of some pretty heinous ways it's being used that a lot of people are not. (APPRISS, which is with zero competition for the title the most evil use of machine learning I have ever seen, and which is probably being used on you right now.)

You need to stop and actually think about why people do bad things with it instead of falling for the red herring and going after the technology (as well as the weakest human target you can find) every time you see those two letters together.

You cannot protect yourself and other people against its misuse if you cannot separate that misuse against its neutral or helpful uses, or if you cannot even identify what AI and machine learning are.

[ultraviolet-divergence]

Perhaps we simply shouldn’t create DAOs (decentralized autonomous organizations, a combination of decentralized finance instruments and smart contracts running on blockchains). Perhaps the legal fiction of corporate personhood is absurd enough without giving the corporations the ability to run themselves. Perhaps we can simply avoid prompt injection attacks against autonomous corporate instruments by not building the torment nexus autonomous corporate instruments.

[yellowjestertfs]

Just another ordinary day

Been busy working on a longer project (as in 40k+ words :0) but in the meantime decided to publish another older story of mine with revisions and images. AI was being especially tricky on me this time so the images are not quite what I pictured but good enough. If anyone has any tips for making better images or is interested in proofreading my longer story let me know!

I woke with a start, my mind still groggy from sleep my vision hazy. It was one of those sudden wake-ups that throws off your whole day, the kind usually prompted by some bad dream or loud noise. Only there had been no such occurrence; my sleep had been peaceful and from what I could remember dreamless, yet I couldn’t shake the feeling something had woken me. 

No matter the cause I was up, and judging by the daylight creeping through my shades there was no point falling back to sleep. With a groan, I lifted myself out of bed and made my way to the bathroom. The alarm on my bedside table informed me I had thirty extra minutes this morning to get ready for work. Never one to waste time I decided to have a quick wank with my extra time to try to release some of the stress my sudden wakeup had caused.

Something felt off as I pulled down my pants to reveal my dick, rock hard as it was most mornings. The type of feeling you get when you say a word over and over and it loses all meaning. Everything else seemed normal, my body was still just as average as when I went to bed, nice strong legs from a childhood of playing soccer and a slight beer belly from my time playing beer pong in college. My face looked the same as well, a generally generic face, adorned by light stubble which had grown in while I slept, and bags under my eyes from my draining corporate job. It was my dick that felt off, foreign, only that was ridiculous. It looked the same as it had since I finished puberty. Just over a foot long and proportionally thick, it was just as average as the rest of my body. Something about thinking of my third leg as average felt wrong but I chalked it up to the dregs of sleep. That was simply how men were, nothing strange about it.

Shanking myself out of my contemplative state I hopped into the shower and went about the act of washing away the sheen of sweat I had gained while I slept. I also took this time to rub one out, using the standard two-hand technique practiced by most men. My dick quickly rose to its full size, and within minutes, my tennis ball-sized balls were churning out cum. I thought back to an article I had read in high school that claimed the average male ejaculated a third a gallon of cum per climax, and judging by my admissions that seemed plausible. I supposed the amount coupled with the force accounted for the high rate of condom breakage, not that any but the bravest of women ever allowed for penetrative sex.

After maneuvering the shower head to force all the cum down the drain I turned off the water and wrapped a towel around my waist, paying special care to ensure that my dick didn’t cause the cloth to come undone. Suddenly I felt a wave pass over me. I felt immediately nauseous and light-headed and a strange sensation of deja vu. I realized this was the feeling that had woken me up this morning, then just as suddenly as it had come over me the queasiness vanished as did my memory of the event. I was left only with a vague sense of unease. Powering through the strange sensation I wiped down the mirror and was confronted once again with a visage that felt somehow off. It wasn’t my average face nor the obscene bulge hidden behind my towel, both of those were normal. My body too looked just as average as ever, thick cut pecs, prominent square abs, and bulging 22’’ biceps were nothing to write home about, although I supposed my time playing soccer had given my legs an extra boost elevating them from the standard 30-inch thickness to a respectable 35. Luckily for me, men are incapable of storing fat otherwise I might have a belly from all those beers I drank in college I thought to myself absentmindedly patting my six-pack. Still, in a world where most men have 250 pounds of walking muscle, I have always felt sort of insecure about my scrawny 230-pound body. 

Quickly forgetting about the strange sensation I finished my morning routine, electing to keep my stubble in the hopes of cultivating a more rugged look on my average face. I exited the bathroom and opened my closet, greeted by the sight of several rows of various dress shirts, embarrassingly all labeled as men's adult small. Putting on underwear was easy enough as with all menswear my boxers had a special compartment for my hose-like junk. A dress shirt too buttoned easily over my cabbage-sized pecs as of course all men's shirts were created for just the task. I was just in the process of squeezing my legs into billowing trousers when I felt another wave pass over me. My already precarious balance caused me to fall, and I caught myself on the edge of my dresser, only it wasn’t a dresser. Why would I have a dresser, I wasn’t a woman what would I do with clothing? Righting myself against what I realized was a workout bench I glanced down just to reassure myself of my nakedness. I wondered absently where the thought of me owning clothing had come from, what a preposterous idea, that would be like a woman walking around naked. I would be fired on the spot if I showed up in such an offensive garment. Casting the ridiculous idea out of my mind I grabbed my bag and headed off to work. 

Saying hello to my hunky neighbor as I passed I finally emerged onto the street. Despite my strange morning, the world outside my apartment appeared the same as it always was, men on their way to work naked, of course, pecs and dick bouncing as they walked, bare feet smacking against the smooth temperature-controlled cement. I joined the throngs of men crowding the sidewalks and waited at a crosswalk as men showing flesh drove by, their cars of course made specifically large enough to hold their bulk. I became just another face in the crowd, just another man on his way to work, bodybuilder frame revealed to the wind. The eye easily passed over my foot-long dick, the instrument not nearly long enough to garner any attention. Be they young or old, rich or poor every man was at least 200 pounds of muscle with a shlong to match and of course, all of them were naked, it was simply how the world was, how it had always been. Depending on the subway station I swiped my metro card and made my way to the appropriate train. As the train pulled I was buffeted by yet another wave and was instantly wracked with an intense pulse of nausea which disappeared just as suddenly as it had arrived. 

Releasing I had fallen down, but not knowing why, I stood back up to my full 7’10” hight and saw all around me men doing the same. For a moment the doors to the subway car in front of me looked strange, almost too tall but that didn’t make any sense. They stood just as tall as ever, the standard 9 foot hight, enough to allow most men to enter without hitting their heads. I knew of course that there were rare men who would still have to duck to enter the train car but for the vast majority of men who averaged around 8’0’’, ten feet was more than sufficient. I entered the car and sat down, my bare butt brushing up against the perky ass of a blond man with a round face on one side and a woman in expertly pressed dress slacks and a matching navy blazer on the other. As the train took off another wave stuck. This one merely caused me to clutch my head as a splitting headache appeared and then vanished in a second. The woman next to me was hit harder by the instantly forgotten wave of reality-altering force. Thrown off balance she bounced into my left pec, her head cushioned by the squishy yet firm muscle. Recovering immediately and feeling somewhat confused as to how she ended up pressed against me she apologized and distracted herself by pulling out her phone and flipping to the camera app to ensure her makeup was not smudged. Though the camera was pointed at herself I could see my reflection, my head towering over hers even in my sitting position. 

I certainly wasn’t ugly by any standard but I also wasn’t some model. My chiseled wide jaw was just about as handsome as every other man on the train, although the perfect coating of square stubble that had grown in during the night did lend me a rugged edge. The rest of my features were pretty mundane, clear and pore-less skin, thick square eyebrows and a dimpled wide chin were the default for men, as evidenced by the golden-haired Adonis that sat next to me. Even so, I always liked my piercing bright eyes and high cheekbones even though they were hardly rare in the world.

The blond man sitting next to me with the perfect lantern jaw got up at the next stop. Mine was the one after that. 

I exited the car and ran up the stairs, taking them two at a time to ensure I wasn’t late. My work building looked the same as it always did, with large doors to accommodate male employees and in the lobby a giant bronze statue of a man holding the earth, his body naked and extremely well muscled and hung of course for the sake of realism. Despite my relative scrawniness I still used a male-designated elevator, the female ones not made to handle my weight or height. The several other men in the elevator and I had only made it a few floors before we were subject to one final and seemingly extra powerful shockwave. The weight of the changes enacted easily caused all the men even with their rock-hard muscles to crumple and we collapsed onto each other. My hand somehow ended up gasping the long penis of a 40-year-old accountant with a perfectly maintained salt and pepper beard. For a moment I motioned to let go of his member before reality snapped back in and I remembered my manners. It would be incredibly rude for me to begin a morning grope and not bring him to completion. In fact,  I had already made a major faux pas by not kissing my coworker hello. This error in tact was quickly rectified as the rest of the elevator ride turned into a make-out session. By my floor the sexy accountant I was giving a handjob to reached completion and I took his load as my breakfast. As I left he spanked my ass and stuck his business card between my butt checks. Guess he liked my elevator pitch. 

I went straight to my boss's office as was customary and gave the 350-pound silver fox a quick blow job before he transferred his abnormally large penis into my ass and fucked me while we discussed business. Turns out the reality-warping machine he had invested in had been broken into this morning although as far as anyone could tell no damage had been done nor had the machine been used. 

“Makes sense I told him" In-between moans as he obliterated my prostate. “I imagine we would know if someone were to fuck with reality.” 

My boss clenched his superhumanly wide lantern jaw and straightened up to his full over eight-foot height, both football-sized biceps flexed behind his head. “You're right on that account kid, today is yet another ordinary day.

[atalienart]

@ianwaite replied to your post “Does Al also make you paranoid? Look at this...”:

Why do you assume that AI artists are not interested in their cause? What do you think of digital art?

​There are no Al artists, there are Al prompters. If you've ever been interested in something then you must know that the object of your interest is what really counts. You want to know more about it, you want to fully experience it. You want to be a writer, you sit and write, you develop your skills, find new techniques of writing and telling stories through written words. If you don't want to write, you're not a writer. You can tell stories but that doesn't mean you're a writer. You can have ideas but that doesn't make you a writer. If you tell another person or a machine to write you something you're not a writer. If someone else composes words for you, you're not a writer. You are not a carpenter when you tell a machine to make you a table in a rococo style. You're not interested in carpentry, you just want a table and want people to clap saying "wow, what a beautiful table, you're such an amazing carpenter" and then to buy it. This is what they're interested in. Al prompters are interested either in the technical aspects of Al, or the final "product" that can give them money or a pat on the back, or a short lasting pleasure of seeing their wife's head on a supergirl body. Tell me the difference between prompting and image searching? There is none.

What do I think about digital art? It's great and beautiful. Graphic programs artists use are another medium. Graphic programs don't lead your hand, don't magically pop images like Al. You draw lines with a stylus on tablet the same way you draw with a pencil on paper. Every medium has different properties. With oil paints you can do impasto paintings, watercolours react beautifully with water, Clip Studio allows you do draw on layers. There is no dark magic but Al prompters see a sped up, edited screen recording of a digital art proces and they think the drawing magically appears on screen without a physical input of the artist. They don't see work, just pats and some "product". I love digital art because it allowed me to share my art with other people much easier. I admire digital artists who can create amazing art pieces. I hope they won't give up when Al people constantly steal from them.

[tlbodine]

T.L. Bodine's Zine Club

Wouldn't it be cool to have something to look forward to in the mail every month?

I thought so, too.

We're living in a world that is increasingly trying to replace human labor, human art, even human thought. And as everything around us becomes more and more inundated with AI slop and algorithmically generated bullshit and sanitized corporatized messaging, I've found myself increasingly interested in creating something real.

So, to that end, I am now offering zines as a Patreon reward.

All zines are written, designed, printed, folded, stuffed into envelopes, and mailed by me.

Contents may include:

Original short fiction

Nonfiction/essays/deep dives

Writing advice and instruction

Interactive elements like journaling prompts

Art, collage, photography, and so forth

Bonus notes, stickers, errata, and other lil bits and bobs I find

Anticipate that they'll be horror-flavored.

I'm still new to zine-making, and I'm looking forward to experimenting with different formats and techniques. I'm looking forward to practicing art and collage and multimedia. I can't guarantee glossy perfection with every issue, but I will always deliver something unique that you won't find in any of my online materials.

Subscribe for $5 on my Patreon. I'll also sell the files so you can print them at home if you'd rather buy them as one-offs.

The first zine will be sent in early July, so subscribe now so you don't miss it :)

[domistique]

I honestly don't know what an AI ffs looks like. You don't have to call anyone out but can you tell us why you think smthing is made by AI?

Domistique's Subjective Guide to Identifying AI Fic in the F1 RPF Fandom

Summaries "In the high-octane world of Formula One," "In the high-stakes world of Formula One," "In the [insert descriptor] world of Formula One,"

My biggest "tell" for AI in summaries. It reads like a blurb for a published original novel. Summaries ending with a question framed with two 'options' also ring alarm bells for me now which SUCKS but the same structure is popping up so frequently alongside other indicators of AI that it's the natural conclusion :(.

Look at the establishing paragraphs.

I'm not a writer so I don't know the terminology, but the paragraphs which set the scene and describe the environment the characters are in. Do they always list elements of the setting in a uniform way which seems too formal for a fic? AI doesn't understand 'show, don't tell'. In my experience it tends to exposition the FUCK out of scenes. Fanfiction is built on the assumption that the reader KNOWS about this stuff already. I've seen this countless times across many fics from different authors which proves to Me it's not just someone's style. Establishing paragraphs starting with a descriptive statement, followed by multiple Example clauses listing sensory aspects of the environment. Usually ending with a picture perfect smartass line, often introducing a character or narrative element (tension, fear, attraction, etc) and creating intrigue. ALL THINGS WHICH ARE NORMAL AND GOOD TECHNIQUES IN FICS! but it's the way they KEEP popping up consistently keeping to the exact. same. paragraph structure.

An author posting multiple new 10,000+ word fics every week for many weeks.

If I suspect a fic is written with AI I usually check the account which posted it to see if they are producing an output which human beings are literally not capable of. If someone says they're working full time in the legal field they are not also writing a dissertation of rpf every three days :(.

Sentences with the same rhythm over and over and over again.

Like. "The wind had eased, the moon reflecting like silver in the lake. The soft hum of the car grounded them, sitting there together on the peaceful shore." I just made that up and you can tell because it sucks BUT it is based on a pattern I keep seeing in fic which I suspect to be AI. ONLY when the other indicators are present. Take this one with a grain of salt because it is mainly vibes-based.

This is mostly only applicable when a fic is written in English :(. I think the ability to tell when something is subtly off with a piece of writing is probably linked to the length of time the reader has been immersed in reading in the language, exposure, etc.

I recommend paying attention to your favourite authors' style, quirks, tendencies; the things which are constant across their works and feel comforting and familiar because you know you're in good hands. Then apply that to AI, which in my opinion has a particular style which it ALWAYS follows when prompted to write fanfic. It knows how to create a plot and make it compelling, even emotional, but it doesn't know how to deviate from the formula (yet). It has improved SO much since the first AI fics started popping up. But I really think it will never be completely indistinguishable from the work of a human because humans are loving and caring and silly and take risks and spend their free time writing stories for us to enjoy for no personal gain. No matter the quality of the writing!!! We as readers owe it to them to read critically and attentively and honour their efforts by rewarding them, not a machine.

(biggest disclaimer ever: i am not an expert this is all based on my observations and i may be wrong. if you don't like someone's sentences don't assume they used AI. it's a sense that had to be honed over time and i will be distraught if some poor writer gets harassed because of this so help me god. just. pay attention to patterns. i have many many examples which i am not including in this post because i don't feel comfortable when i can't be 100% sure they are the product of AI. i can dm them if you want!!)

[chriskotiesen]

Just saw and adored the creativity in your creature design work on my fyp

Do u perhaps have any tips for brainstorming ideas? Currently im just going with moodboards on pinterest but its kinda clogged with ai art...

Thank you!

I can't help you with your Pinterest AI problem, but I have plenty of brainstorming techniques. I wrote about my thought process a bit when I walked through one of my designs last year (click here). In general though, I always start my creatures either by just smashing shapes together, or by building off of a vague concept.

STARTING WITH SHAPES

• Draw some random shapes, then try to find creatures in them. If you don't see anything interesting in a shape right away then you can try adding to it or flipping it around. If that doesn't work, just abandon it and draw another.

• If you want to push your creativity a bit further, immediately discard the first idea you come up with and try to find a second way of interpreting the same shape. That will pretty much always get you something a bit stranger and less like your usual work.

STARTING WITH CONCEPTS

• Come up with a really wild design prompt for a creature and then try to make it work. Something like "eyes on its knees", "three heads but only one limb", or "crustacean with feathers".

• Pick a feature of a real animal that you find interesting and do something with that. Nature has all kinds of beautiful and fascinating stuff that you can borrow! Sometimes creature design to me is just about taking some little things that I find neat (like crab legs, or jellyfish oral arms, or goat eyes) and trying to present them in a way that hopefully makes other people see them as interesting too.

• Also consider strange variations of real features. What if bird wings were jointed differently, or if they had something other than feathers laid out in the same pattern? What if a creature with tusks also grew several rows of teeth at once, like a shark?

• Try formulating a concept around a creature's abilities. What would a supersonic bird look like? Or a bug that communicates via radio waves?

• If you want to get really silly or alien, you can try blending in ideas from non-organic sources. What if a typewriter was an animal and all of its little arms were actual limbs? Can you give an animal wheels in a way that looks evolutionarily plausible?

GENERAL CREATURE DESIGN ADVICE

• Not every creature will come out great, and that's okay. I draw A LOT and the majority of my work doesn't get shared online. But the more things you try, the better the odds that you'll find something good.

• You want to strike a good balance between novelty and relatability. Too normal and it won't be very interesting. Too weird and it won't have any personality. Once you have a vague idea, you can often add or remove features to reach the balance that you want.

• No matter how crazy the overall body plan may be, it's the little details that make something feel natural. Do the joints look like they would bend in a sensible way? How does the texture of its skin vary across its body? Can you imagine what petting it would feel like?

Hope that helps!

[tadpole-apocalypse]

I’m trying to mostly stay off Reddit as much as possible these days, but I still participate in some art related subs because I like offering critiques and suggestions from my own experience.

There’s a kind of frustrating trend where someone will post a polished piece of art and ask “what brush can I use to achieve this” and it’s like…nothing to do with the brush, it’s technique, blending and skill that was built up over many years or decades.

I also get these are questions by amateur and beginner artists (if not grifters trying to identify an art style to plug into ai prompts) but there are also so so so many resources for new artists now I just want to shake them by the shoulders and just scream “Experiment with the brushes that came with your program! Just try painting it! Dont get frustrated when it doesn’t look exactly like your reference art! You are still learning! There is no brush that will be a shortcut to a finished rendered piece!”

[canmom]

oh no she's talking about AI some more

to comment more on the latest round of AI big news (guess I do have more to say after all):

chatgpt ghiblification

trying to figure out how far it's actually an advance over the state of the art of finetunes and LoRAs and stuff in image generation? I don't keep up with image generation stuff really, just look at it occasionally and go damn that's all happening then, but there are a lot of finetunes focusing on "Ghibli's style" which get it more or less well. previously on here I commented on an AI video model generation that patterned itself on Ghibli films, and video is a lot harder than static images.

of course 'studio Ghibli style' isn't exactly one thing: there are stylistic commonalities to many of their works and recurring designs, for sure, but there are also details that depend on the specific character designer and film in question in large and small ways (nobody is shooting for My Neighbours the Yamadas with this, but also e.g. Castle in the Sky does not look like Pom Poko does not look like How Do You Live in a number of ways, even if it all recognisably belongs to the same lineage).

the interesting thing about the ghibli ChatGPT generations for me is how well they're able to handle simplification of forms in image-to-image generation, often quite drastically changing the proportions of the people depicted but recognisably maintaining correspondence of details. that sort of stylisation is quite difficult to do well even for humans, and it must reflect quite a high level of abstraction inside the model's latent space. there is also relatively little of the 'oversharpening'/'ringing artefact' look that has been a hallmark of many popular generators - it can do flat colour well.

the big touted feature is its ability to place text in images very accurately. this is undeniably impressive, although OpenAI themeselves admit it breaks down beyond a certain point, creating strange images which start out with plausible, clean text and then it gradually turns into AI nonsense. it's really weird! I thought text would go from 'unsolved' to 'completely solved' or 'randomly works or doesn't work' - instead, here it feels sort of like the model has a certain limited 'pipeline' for handling text in images, but when the amount of text overloads that bandwidth, the rest of the image has to make do with vague text-like shapes! maybe the techniques from that anthropic thought-probing paper might shed some light on how information flows through the model.

similarly the model also has a limit of scene complexity. it can only handle a certain number of objects (10-20, they say) before it starts getting confused and losing track of details.

as before when they first wired up Dall-E to ChatGPT, it also simply makes prompting a lot simpler. you don't have to fuck around with LoRAs and obtuse strings of words, you just talk to the most popular LLM and ask it to perform a modification in natural language: the whole process is once again black-boxed but you can tell it in natural language to make changes. it's a poor level of control compared to what artists are used to, but it's still huge for ordinary people, and of course there's nothing stopping you popping the output into an editor to do your own editing.

not sure the architecture they're using in this version, if ChatGPT is able to reason about image data in the same space as language data or if it's still calling a separate image model... need to look that up.

openAI's own claim is:

We trained our models on the joint distribution of online images and text, learning not just how images relate to language, but how they relate to each other. Combined with aggressive post-training, the resulting model has surprising visual fluency, capable of generating images that are useful, consistent, and context-aware.

that's kind of vague. not sure what architecture that implies. people are talking about 'multimodal generation' so maybe it is doing it all in one model? though I'm not exactly sure how the inputs and outputs would be wired in that case.

anyway, as far as complex scene understanding: per the link they've cracked the 'horse riding an astronaut' gotcha, they can do 'full glass of wine' at least some of the time but not so much in combination with other stuff, and they can't do accurate clock faces still.

normal sentences that we write in 2025.

it sounds like we've moved well beyond using tools like CLIP to classify images, and I suspect that glaze/nightshade are already obsolete, if they ever worked to begin with. (would need to test to find out).

all that said, I believe ChatGPT's image generator had been behind the times for quite a long time, so it probably feels like a bigger jump for regular ChatGPT users than the people most hooked into the AI image generator scene.

of course, in all the hubbub, we've also already seen the white house jump on the trend in a suitably appalling way, continuing the current era of smirking fascist political spectacle by making a ghiblified image of a crying woman being deported over drugs charges. (not gonna link that shit, you can find it if you really want to.) it's par for the course; the cruel provocation is exactly the point, which makes it hard to find the right tone to respond. I think that sort of use, though inevitable, is far more of a direct insult to the artists at Ghibli than merely creating a machine that imitates their work. (though they may feel differently! as yet no response from Studio Ghibli's official media. I'd hate to be the person who has to explain what's going on to Miyazaki.)

google make number go up

besides all that, apparently google deepmind's latest gemini model is really powerful at reasoning, and also notably cheaper to run, surpassing DeepSeek R1 on the performance/cost ratio front. when DeepSeek did this, it crashed the stock market. when Google did... crickets, only the real AI nerds who stare at benchmarks a lot seem to have noticed. I remember when Google releases (AlphaGo etc.) were huge news, but somehow the vibes aren't there anymore! it's weird.

I actually saw an ad for google phones with Gemini in the cinema when i went to see Gundam last week. they showed a variety of people asking it various questions with a voice model, notably including a question on astrology lmao. Naturally, in the video, the phone model responded with some claims about people with whatever sign it was. Which is a pretty apt demonstration of the chameleon-like nature of LLMs: if you ask it a question about astrology phrased in a way that implies that you believe in astrology, it will tell you what seems to be a natural response, namely what an astrologer would say. If you ask if there is any scientific basis for belief in astrology, it would probably tell you that there isn't.

In fact, let's try it on DeepSeek R1... I ask an astrological question, got an astrological answer with a really softballed disclaimer:

Individual personalities vary based on numerous factors beyond sun signs, such as upbringing and personal experiences. Astrology serves as a tool for self-reflection, not a deterministic framework.

Ask if there's any scientific basis for astrology, and indeed it gives you a good list of reasons why astrology is bullshit, bringing up the usual suspects (Barnum statements etc.). And of course, if I then explain the experiment and prompt it to talk about whether LLMs should correct users with scientific information when they ask about pseudoscientific questions, it generates a reasonable-sounding discussion about how you could use reinforcement learning to encourage models to focus on scientific answers instead, and how that could be gently presented to the user.

I wondered if I'd asked it instead to talk about different epistemic regimes and come up with reasons why LLMs should take astrology into account in their guidance. However, this attempt didn't work so well - it started spontaneously bringing up the science side. It was able to observe how the framing of my question with words like 'benefit', 'useful' and 'LLM' made that response more likely. So LLMs infer a lot of context from framing and shape their simulacra accordingly. Don't think that's quite the message that Google had in mind in their ad though.

I asked Gemini 2.0 Flash Thinking (the small free Gemini variant with a reasoning mode) the same questions and its answers fell along similar lines, although rather more dry.

So yeah, returning to the ad - I feel like, even as the models get startlingly more powerful month by month, the companies still struggle to know how to get across to people what the big deal is, or why you might want to prefer one model over another, or how the new LLM-powered chatbots are different from oldschool assistants like Siri (which could probably answer most of the questions in the Google ad, but not hold a longform conversation about it).

some general comments

The hype around ChatGPT's new update is mostly in its use as a toy - the funny stylistic clash it can create between the soft cartoony "Ghibli style" and serious historical photos. Is that really something a lot of people would spend an expensive subscription to access? Probably not. On the other hand, their programming abilities are increasingly catching on.

But I also feel like a lot of people are still stuck on old models of 'what AI is and how it works' - stochastic parrots, collage machines etc. - that are increasingly falling short of the more complex behaviours the models can perform, now prediction combines with reinforcement learning and self-play and other methods like that. Models are still very 'spiky' - superhumanly good at some things and laughably terrible at others - but every so often the researchers fill in some gaps between the spikes. And then we poke around and find some new ones, until they fill those too.

I always tried to resist 'AI will never be able to...' type statements, because that's just setting yourself up to look ridiculous. But I will readily admit, this is all happening way faster than I thought it would. I still do think this generation of AI will reach some limit, but genuinely I don't know when, or how good it will be at saturation. A lot of predicted 'walls' are falling.

My anticipation is that there's still a long way to go before this tops out. And I base that less on the general sense that scale will solve everything magically, and more on the intense feedback loop of human activity that has accumulated around this whole thing. As soon as someone proves that something is possible, that it works, we can't resist poking at it. Since we have a century or more of science fiction priming us on dreams/nightmares of AI, as soon as something comes along that feels like it might deliver on the promise, we have to find out. It's irresistable.

AI researchers are frequently said to place weirdly high probabilities on 'P(doom)', that AI research will wipe out the human species. You see letters calling for an AI pause, or papers saying 'agentic models should not be developed'. But I don't know how many have actually quit the field based on this belief that their research is dangerous. No, they just get a nice job doing 'safety' research. It's really fucking hard to figure out where this is actually going, when behind the eyes of everyone who predicts it, you can see a decade of LessWrong discussions framing their thoughts and you can see that their major concern is control over the light cone or something.