I've been doing a lot of analysis — rather, letting AI do the objective analysis of certain texts — and in the process realized that ChatGPT can read/"see" screencaps. So I thought (given the fucking constant flood of links I've begged you cunts not to send), wouldn't it be fun to run the Twitter Bitches through AI to see what it says about them?
Don't even start. I have no time to talk to AI about this shit, so this was the first time I've done so, i.e. this AI of mine doesn't have any clue to my background or dealings with the Twitter bitches. There's no history or prior programming.
Starting with EIE (since she can never shut the goddamn fuck up about Percy White), I uploaded screens of her shit and asked it this: "Analyze this person's posts in terms of evidence and/or motivations towards Percy Hynes White." I did not feed it any of my own sentiment about the case at all and remained neutral when asking what it could do (i.e., I asked if it could read a thread and check for its veracity and judgement, and it said yes).
These are the results. I have lots of screencaps for this, so I'm skipping to its final initial judgement in the beginning regarding the pinned accusations thread:
But then I realized something that was deficient in the original analysis: His age was absent from consideration (yeah yeah, it spelled his name wrong, I was on voice-to-text):
Accurate. You can do it yourself, with your own screencaps, however I'm not sure I trust any of the Twitter Bitches to go in neutral.
Anyway, I sent in the T-Shirt Debacle Tweets (LOL) after because I was curious what it would assess from its collective Tweets provided:
ETA: I added "Tumblr Mom" to the title because, direct quote: "She behaves like a "Tumblr mom" — a woman in her 30s or very late 20s who came of age in a callout-heavy, fandom-meets-activism online environment." 💀💀💀💀💀
Trust. It's the same with the niall/1D stan reject.
Anya is live and ready to show you everything. Watch her strip, dance, and perform exclusive shows just for you. Interact in real-time and make your fantasies come true.
✓ Live Streaming✓ Interactive Chat✓ Private Shows✓ HD Quality✓ Free Actions
Free to watch • No registration required • HD streaming
Fandom space has always thrived on the creativity and collaboration of fans. From AUs to crackfics, creativity is what makes the fandom world go around. Is it really a surprise, then, that A.I. (artificial intelligence) and LLMs (large language models) are often looked down on in such spaces? A robot that predicts the most likely next word in a sentence isn't creative; it's the antithesis of everything fandom culture has stood for. All of a sudden, people who had been writing for their own creative fulfilment had their work stolen, used to train A.I. models to generate profit. Fandom became a place hostile to A.I. as soon as A.I. entered it, but this hostility has led to fear among writers and authors alike.
How do I sound human?
You write. Seriously, begin writing.
I hate to sound cheesy, but you are the only thing holding yourself back in this instance. If you're sat reading this section, you're afraid of judgement and accusation; you're so afraid of being accused of something that you so vehemently disagree with that you won't even try what you love. Isn't that sad? No matter what you do to avoid sounding like A.I., those tips will be taken by other writers, those writers will publish their work, whether traditionally or online, and that work will be scraped as training data. Any advice that you get, no matter how much you avoid "A.I. tells" like emdashes or tildes, will become redundant in a matter of months as new works get fed to the machine and a new set of "tells" is discovered.
Use tildes, use semicolons and oxford commas. Ignore those who make baseless accusations, you can't tell whether something is human-written or not by its vibe. A.I. learned to write from free content on the internet. Youtube comments, reddit threads, and perhaps most importantly, fanfiction websites like wattpad, ao3, and fanfiction.net. No matter what you write, you will never "sound like" A.I., because A.I. sounds like you, it sounds like us, the fandom collective.
Aside from that, find your style in writing and be confident in yourself. One of my favourite lines in fiction is "the ships hung in the sky in much the same way that bricks don't", from Douglas Adams's A hitchhiker's guide to the galaxy. Douglas Adams in general has a very distinct, very humorous writing style. He is unabashedly silly, but even his jokes' setup is a witty device. Many people, myself included, managed a giggle out of the imagery of a brick hitting the ground with an unceremonious crack, but there is also merit in the metaphor itself. Metaphors naturally exist as a comparison, by juxtaposing the idea of the ships and a brick, it brings to mind the size of the ships, their impossible weight, and how unnatural it is for them to be flying in the first place. This is not the only example of his distinct humour hiding other messages, such as "Nothing happened. And then, nothing continued to happen." cementing the finality of death, or "It's unpleasantly like being drunk", "What's so unpleasant about being drunk?", "Ask a glass of water." subverting the expectation that drunk refers to being intoxicated, when in reality it is referring to the act of drinking fluids at all, and illustrating the danger the characters are in. All in all, a witty writing style that allows the reader to lower their guard while painting a vivid picture at the same time.
I am not saying to mimic the writing style of Douglas Adams. For as much as I like it personally, it's his, and it should remain that way. Do, however, take inspiration from writing styles you enjoy.
How do I protect myself?
Honestly, there isn't a ton you can do, but that doesn't mean there is nothing. Hiding your work is probably safest. If you're posting to AO3, like I do, click the "only show work to registered users" box. The harder it is for A.I. to access your work, the less likely that work is to be scraped. There is no poisoning LLMs in the same way that one can poison an image. There is, however, something to be said about how this discourages fandom practice and etiquette. Fandom exists to be shared, hundreds or thousands of people coming together to share their interpretation of the story, how they feel a character should or shouldn't act according to their interpretation. The harder it is for A.I. to access your work, the harder it is for fans to do the same. I can't count the amount of fics or AUs I've personally abandoned because viewing required an account on a site I didn't use. A.I. based paranoia, as much as it is justified, rips fandom apart from the inside.
Until laws change, be vocal. Join protests, online or in person, sign petitions. Do not engage in witch hunts, it harms fandom and it fosters distrust.
Turn to writing software that is either offline or vehemently against generative A.I., such as ellipsus. They actually sell merch to fund their development, such as their "defend the em-dash" stickers, "resistance starts with words" shirt, or "this machine kills fascists" typewriter stickers. On top of that, their "plus" model is based on aesthetics along with showing your own personal statistics and operates on a "subscribe to own" model, meaning that if you've subscribed for 24 months (continuous or not), you no longer have to pay. In case it isn't clear, this is the software I personally use after I abandoned google docs.
What if I use AI?
Then that is your choice. I disagree with it, but what lines are drawn are up to the individual. All I ask is that you disclose it. Not doing so, you're betraying your readers, and making them engage in content they would otherwise avoid. AO3 even has a tag for AI-generated works! The best example I could give for this, if someone with a peanut allergy orders chips/fries at a restaurant, can you blame them for getting upset when they have an allergic reaction? No, you would instead get mad at the restaurant for not disclosing that their fryers use peanut oil. In the same way that it's reasonable to assume the fryers use vegetable oils, it's reasonable to assume fanfic, a section of the internet based on human creativity and passion, is made by humans.
A take away
A.I. usage is looked down on in fandom, I believe the reasons for this are reasonable. There have been studies documenting the negative effects of A.I. generated texts on authors, and how it saps creativity. Hunting for people who may or may not be using A.I. sows distrust in a space that has always been collaborative by nature, we wouldn't have omegaverse if not for team world building. If you use A.I., please disclose it, so people can make informed choices about whether or not to engage in your work. Consent remains important, informed consent even more so.
A good way to find out whether a piece of art is AI is to search for logical errors. Human artists add everything with a purpose even if they might lack the skill for it, but AI doesn't understand what it's doing since it's only copying what it sees (note: you can see a similar effect when people trace art, but don't understand what it is they're tracing)
An artist might make a logical error or two just because art has many details that you need to keep track of over the course of a few days, but most of the picture has consistency. A major point is to search for as many inconsistencies as possible because they all add up to prove that the art is AI
But enough of that, allow me to give you a visual example using this AI picture of Hatsune Miku:
As with all AI art, it looks fine at a glance which is why you have to look deeper to find your proof
Starting out with the most well known sign of AI use: the hands
The first hand is a mangled mess, while the second one has an extra long finger that turns into hair. This alone can prove that the art is AI (though do be cautious, artists sometimes miscount fingers in their art! If there are more or less than 5 fingers but they all look real, look for other signs in the art before accusing of AI)
Another thing to keep in mind is that AI is terrible at continuation. That is to say, if there is something intersecting where an object goes, it will not understand that the object behind it is the same one. An example in this picture would be the pigtail that's separated by the arm:
Note how thin the pigtail is at the top, but the bottom comes out much thicker as if she grew extra hair. This effect is much more obvious in other pictures so you may call this nitpicking, but the point is to show you methods & what to look for
Another example would be the legs and the skirt:
The leg is coming from in front of the skirt??
One more way to easily tell AI apart is the eyes, as the eyes can be heavily stylized, but AI has troubles keeping them looking consistent with each other:
Miku's left eye is noticeably lighter and more blue (no, it's not the shading when her entire face is supposedly under the same shadow), the edge of the iris is misshapen, there are many random shapes in the eyes. The pupil is probably supposed to be the big lighter circle in the middle
Miku's right eye has less shapes in it and the pupil seems to be a tiny light dot in the middle?
The highlights in the eyes are also done differently and inconsistently. Usually even if the eyes are drawn quite different, the highlights are what makes them come together, even in anime where the focus is on animation & not so much consistency
These are the main inconsistencies to look out for in any art, but even that might not be enough at times, especially if a person is drawing over AI art to try and fix mistakes. Unfortunately, I have seen it before and it is something to look out for, so let's quickly point out any other logical mistakes in this Hatsune Miku picture so you have a better understanding of what to look out for:
The shape of this object is square, but there's a weird angle at the top
2. The pink hood is not attached to anything on one side. Additionally, her hair blends into the hoodie string
3. Bottom hair strand isn't attached to the rest of the hair (the hair is coming from the top right, yet it ends there. Unless a hair strand did a 180 turn behind Miku, it has no purpose looking like that)
4. The pink hoodie has a single button hole in the middle of one side, and one at the bottom of the other side. This is also odd considering it's supposed to be a hoodie, which doesn't usually have buttons. Additionally, I'm sure the text of the shirt is messed up and is unreadable in any language
5. Inconsistent hoodie colors, the left side has a white bottom while the right side has a pink edge. Also it splits into two on the right side for some reason. Also also, the left side has something that looks like a messed up pocket with pink rocks attached to it, which doesn't make sense when the bottom of the hoodie has a different type of pocket.
6. The background. Note the railing starts right above the yellow on the left side, in the middle of the bridge. But then suddenly is at the far edge of the bridge on the right side with the white road between the two
Lastly for good measure, take a look at the whole picture again and look at how inconsistent the shading is:
The lighting can be split into two parts, where the light comes from the top of the character & where the light comes from the right side
Light comes from the top:
Top of the hair
Raised arm
Top of the hoodie & shirt
Skirt
Light comes from the right side:
Pigtails
Bottom of the jacket
Fluffy collar
Face
You can't justify it by saying "it's coming from the top right" because that's not how the shading would work in that situation. If it was, the lowered arm wouldn't have such a bright light since the light wouldn't be able to reach it. Note that the raised arm also doesn't cast a shadow over the chest
A lot of these things are something you understand through experience in art, so it's okay if you struggle to see what I described here. After all, I'm here to help so that I can use my artistic experience to better identify AI art as well as help you learn the signs of it
AI art is getting harder to spot which can trick you into supporting so-called 'AI artists'
To avoid that, you can anonymously send any suspicious looking art and/or accounts here and I'll analyze the artwork to see if I spot any traces of AI use. Hopefully over time we can all learn to tell AI art easier so we don't fall victim to its lies
Keep in mind that at the end of the day I'm just one person doing my best. I'm an artist who's analyzed plenty of AI art so I've gotten a hang of telling AI art apart, but I can make mistakes especially as they try to cover their tracks more and more
Anya is live and ready to show you everything. Watch her strip, dance, and perform exclusive shows just for you. Interact in real-time and make your fantasies come true.
✓ Live Streaming✓ Interactive Chat✓ Private Shows✓ HD Quality✓ Free Actions
Free to watch • No registration required • HD streaming
AI Energy Consumption Market Trends and Industry Analysis
The AI vs Energy Consumption: Long-Term Grid Impact Study shows strong AI Energy Consumption Market Trends driven by increasing reliance on artificial intelligence across industries such as healthcare, finance, and transportation.
AI-powered systems are now widely used in automation, cloud services, and predictive analytics. These systems require high-performance computing infrastructure, which leads to rising electricity consumption globally.
The expansion of data centers is one of the major factors contributing to increased energy usage. As companies scale their AI operations, energy demand continues to grow rapidly.
Artificial intelligence has swiftly evolved from a specialized computational area into a dominant force driving electricity demand growth in the 21st century. This trend is discussed in the AI Energy Consumption Industry Analysis available at AI vs Energy Consumption Market Trends.
Industry leaders are now focusing on sustainable AI development strategies. This includes energy-efficient chips, improved cooling systems, and green data center initiatives.
Overall, AI Energy Consumption Market Trends indicate strong growth but also highlight the urgent need for balancing innovation with environmental responsibility.
Over the past year and a half, I have spent countless hours interacting with ChatGPT for everything from technical troubleshooting and historical research to memoir writing, civic analysis, and creative experiments. I started out slow and simple, asking to summarize session notes from a Dungeons & Dragons campaign.
Today, I was at a writing salon. We talked a bit about ChatGPT, how it could be…
