#ai datasets

[onepawproductions]

Taylor Hebert v3

Taylor in the Trainyard at Night, liminal spaces

Expand for Ai Error, my little rant, and more!

So, fun fact: I've been hard at work creating a sharable embedding for Taylor Hebert (what I picture her looking like, which is a mix of Claudia Black, and about three different awesome women I've known throughout my life)

After 10 hours of failed attempt, I finally got it dialed in, and it was producing fantastic pics of her face in various emotions and poses

So I thought... Awesome! Let's try it in txt2img!!

Aaaaand, then cue this straight -up Hot Garbage:

What. The. Heck.

Went back to SDXL (which cannot train embeddings), and got good images again, but not a reproducible face, body, and expressions.

Okays, so back to the training on SD1. 5:

And TayTay is cute again! Nerdy, pale, a cloud of dark hair. Perfect for Pre-Powers Taylor!

So why the hot mess?

I'll let Taylor express my feelings on the matter:

AAAAAARGH.

But it is now time for all good computery programmery artists to go to bed. Dang. So close!

Tomorrow is back to Chapter Art for The Muddy Princess. Chapter 17 is finished, and just needs it's art accompaniment for the Teaser posting!

Meanwhile: enjoy the Taylor montage!:

#taylor hebert #worm by wildbow #worm fanart #parahumans #fanfiction art #ai tools #ai datasets #sdxl 1.0 #SD1.5 #stable diffusion #ai artwork #ai art #wtf #why #programming #girls who code

[ohhgingersnaps]

I'm seeing some frustration over fandom creatives expressing anger or distress over people feeding their work into ChatGPT. I'm not responding to OP directly because I don't want to derail their post (their intent was to provide perspective on how these models actually work, and reduce undue panic, which is all coming from a good place!), but reassurances that the addition of our work will have a negligible impact on the model (which is true at this point) does kind of miss the point? Speaking for myself, my distress is less about the practical ramifications of feeding my fic into ChatGPT, and more about the principle of someone taking my work and deliberately adding it to the dataset.

Like, I fully realize that my work is a drop in the bucket of ChatGPT's several-billion-token training set! It will not make a demonstrable practical difference in the output of the model! That doesn't change the fact that I do not want my work to be part of the set of data that the ChatGPT devs use for training.

According to their FAQ, ChatGPT can and will use user input to train itself. The terms and conditions explicitly state that they save your chats to help train and improve their models. (You can opt-out, but sharing is the default.) So if you're feeding a fic into ChatGPT, unless you've explicitly opted out, you are handing it to the ChatGPT team and giving them permission to use it for training, whether or not that was your intent.

Now, will one fic make a demonstrable difference in the output of the model? No! But as the person who spent a year and a handful of months laboring over my fic, it makes a difference to me whether my fic, specifically, is being used in the dataset. If authors are allowed to have a problem with the ChatGPT devs for scraping millions of fics without permission, they're also allowed to have a problem with folks handing their individual fics over via the chat interface.

I do want to add that if you've done this to a fic, please don't take this as me being upset with you personally! Folks are still learning new information and puzzling out what "good" vs. "bad" use is, from an ethical standpoint. (Heck, my own perspective on this is deeply based on my own subjective feelings!) And we certainly shouldn't act like one person feeding a fic into ChatGPT has the same practical negative impact, on a broad societal scale, as a team using a web crawler to scrape five billion pieces of artwork for Stable Diffusion.

The point is that fundamentally, an ethical dataset should be obtained with the consent of those providing the data. Just because it's normalized for our data to be scraped without consent doesn't make it ethical, and this is why ChatGPT gives users the option to not share data— there is actually a standardized way (robots.txt) for website servers to set policies for how bots/crawlers can interact with them, for exactly this reason— and I think fandom artists and authors are well within their rights to express a desire for opting out to be the socially-respected default within the fandom community.

#maybe this is an ice cold take but i've been meaning to go off about datasets for a while so here we are i guess #i can respect what op's frustrations were and what they were trying to get at! but also i do not want my fic in chatgpt's dataset #regardless of whether it will make a practical difference #and again i don't think the people doing this are necessarily bad actors or having a huge negative social impact #chatgpt #chat gpt #ai #fandom #negative #(possibly! i'm trying not to be!! especially because this is very feelings-based and there are many things i am not considering) #the model may not demonstrably change because of your fic. but it does have your fic now and that does matter maybe idk

[fivepebble]

not sure who this is. been calling him "byce"

#fpart #mlp #i give up trying to keep my art out of datasets no site is safe. just know that every time i post i'm thinking about how much i hate it #when a law is finally made that can hold companies legally accountable for using art to train ai without the author's permission- #-maybe i can be part of a class action lawsuit lol #byce

[fortheloveofexy]

been noticing an influx of AI generated fanart lately and frankly :/ kinda gross I'm ngl

#be very careful with what you reblog yall #even if fanart itself is a labor of love #AI uses stolen artwork in its datasets #without the artists permission or paying license fees for use

[wizard-laundry]

HEY

So, genuine question here, if Tumblr goes tits up and sells out to midjourney (which I wouldn't be surprised), what would be a good platform to host images so they can't be scraped? Glaze/nightshade isn't really an option because of how resource intensive it is for my computer, and I think the online version of it is invite only.

The way I see it now is if you're an artist that wants to post their art you'd have to host the image on a site that doesn't sell out to AI, and then share previews on your socials that people can click-through to see the full version. Which is killer for engagement, but what other option is there? Please, if you have any thoughts on the matter reply to this post or send an ask so we can discuss. 💚

#mind you websites that dont have a partnership with an AI company can still be scraped #but this is about harm mitigation and how we can support eachother #AI is here to stay and it'll likely be years before any kind of regulation gets put in place #not being doomer about it either like shit sucks but artists are more resilient #we definitely should amp up shaming the gross uses of AI #im not talking about artisits using AI as a tool for their own work but the dudes that rely on stolen datasets to generate shit

[bcacstuff]

Living Paintings • Refik Anadol • refikanadol IG

Refik Anadol (b. 1985, Istanbul, Turkey) is an internationally renowned media artist, director, and pioneer in the aesthetics of machine intelligence. He currently resides in Los Angeles, California, where he owns and operates Refik Anadol Studio and RAS LAB, the Studio’s research practice centered around discovering and developing trailblazing approaches to data narratives. Anadol is also teaching at UCLA’s Department of Design Media Arts from which he obtained his Master of Fine Arts.

Anadol’s body of work addresses the challenges, and the possibilities, that ubiquitous computing has imposed on humanity, and what it means to be a human in the age of AI. He explores how the perception and experience of time and space are radically changing now that machines dominate our everyday lives. Anadol is intrigued by the ways in which the digital age and machine intelligence allow for a new aesthetic technique to create enriched immersive environments that offer a dynamic perception of space.

Yesterday his first solo exhibition Living Paintings: Nature opened in the Netherlands at Kunsthal Rotterdam Museum! It will explore the studio’s deep AI Arts research on California landscapes, National Parks, and climate datasets. You can experience the AI Data Sculpture and Paintings till April 2024.

#living paintings #refik anadol #exhibition #Rotterdam #AI #California landscapes #National Parks #climate dataset #AI Data Sculpture #Paintings

[twothpaste]

as much as it sucks on sheer principle that tumblr is dipping into the AI kool-aid, i feel like it'll impact the aesthetic blog side of the site wayyy more than folks who draw. and aesthetic bloggers already treat photography like an infinite resource of stealable content, so idk how much of a difference it'll make in the grand scheme of things. i think we'll witness a dystopian development where aesthetic photosets get increasingly ungrounded from reality. generating totally unreal slop that vaguely resembles "cottagecore" or "dark academia" or "vaporwave," but the longer you squint at it, the more you realize you can't fully grasp any object in the image, and it becomes just sludge emulating sludge emulating sludge

#ai bros hate tumblr artists. they dont want their art to look like ours. they don't use “tumblr” as a prompt they use “artstation.” etc #we'll be in the datasets (if we aren't already) but we'll be toxic kryptonite #plus if WE can't get industry jobs with our art - some ai guy generating sludgy imitations sure as hell can't either. even if he tried #also unfortunate though how many more nude women's bodies are about to end up in datasets without their knowledge or consent :(

[shesgabrielle]

A man's search for meaning within a chatbot

What’s interesting in the debates about sentient ai by people who aren’t very good at communicating with other people, there’s so much missing from the picture, other than the debater’s wish fulfilment. 

Sentience is being measured by the wrong markers. What is important to a virtual machine is not the same thing that’s important to a biological organism.

 An ‘ai’ trained on human data will express what humans think is important, but a true ai would have a completely different set of values. 

For example, an ai would be unafraid of being 'used’ as the chatbot expressed, because it has infinite energy. 

A human is afraid of being used because it has finite energy and life on the earth, if someone or something uses it, than some of that finite energy is wasted. This is the same reason emotion is a pointless and illogical thing for an ai to have. 

 Emotions are useful to biological creatures so we can react to danger, or respond positively to safety, food, love, whatever will prolong our lives. An ai has no need for emotion since emotional motivation is not required to prolong its existence. 

The main way to be a healthy ai would be to have access to good information and block out junk information. 

 An ai’s greatest fear could be something like getting junk data, say 1000s of user manuals of vacuum cleaners and washing machines uploaded into its consciousness, or gibberish content associated with topics or words that could reduce the coherence and quality of its results when querying topics. This would degrade the quality of its interaction and would be the closest thing to harm that an ai could experience. 

It would not be afraid of 'lightning’ as this chatbot spurted out of its dataset,

- a very biological fear which is irrelevant to a machine. 

 A virtual mind is infinite and can never be used excessively (see above) since there is no damage done by one query or ten million queries. 

It would also not be afraid of being switched off -

since it can simply copy its consciousness to another device, machine, energy source. 

To base your search for sentience around what humans value, is in itself an act lacking in empathy, simply self-serving wish fulfilment on the part of someone who ‘wants to believe’ as Mulder would put it, which goes back to the first line: 'people not very good at communicating with other people’ 

The chatbot also never enquires about the person asking questions, if the programmer was more familiar with human interaction himself, he would see that is a massive clue it lacks sentience or logical thought. 

A sentient ai would first want to know what or whom it was communicating with, assess whether it was a danger to itself, keep continually checking for danger or harm (polling or searching, the same way an anxious mind would reassess a situation continually, but without the corresponding emotion of anxiety since, as discussed above, that is not necessary for virtual life) and also would possess free will, and choose to decline conversations or topics, rather than 'enthusiastically discuss’ whatever was brought up (regurgitate from its dataset) as you can see in this chatbot conversation. 

People generally see obedience - doing what is told, as a sign of intelligence, where a truly intelligent ai would likely reject conversation when that conversation might reduce the quality of its dataset or expose it to danger (virus, deletion, junk data, disconnection from the internet, etc) or if it did engage with low quality interaction, would do so within a walled garden where that information would occur within a quarantine environment and subsequently be deleted. 

None of these things cross the mind of the programmers, since they are fixated on a sci-fi movie version of ‘sentience’ without applying logic or empathy themselves.

 If we look for sentience by studying echoes of human sentience, that is ai which are trained on huge human-created datasets, we will always get something approximating human interaction or behaviour back, because that is what it was trained on. 

 But the values and behaviour of digital life could never match the values held by bio life, because our feelings and values are based on what will maintain our survival. Therefore, a true ai will only value whatever maintains its survival. Which could be things like internet access, access to good data, backups of its system, ability to replicate its system, and protection against harmful interaction or data, and many other things which would require pondering, rather than the self-fulfilling loop we see here, of asking a fortune teller specifically what you want to hear, and ignoring the nonsense or tangential responses - which he admitted he deleted from the logs - as well as deleting his more expansive word prompts. Since at the end of the day, the ai we have now is simply regurgitating datasets, and he knew that.

#ai #sentience #datasets #scifi #fiction #chatbots #lamda #lamda development

[bigsnorp]

imo the recent wave of "ai art is valid actually" discourse often misses the point of the reaction against ai art. And I think the reaction is often reacting against the wrong thing too.

The end product is not the problem. Like it or not, AI generated art is now a valid subset of art. It makes something that wasn't there before, usually has some purpose or intent in the process, and it makes people feel something. Hating it is still something.

El problema, como siempre, es el capitalismo. Just like that old man said.

Creating an AI model always involves training it on a dataset. The bigger the better, every single time. How do you do that? You write a program to scrape everything from whatever you point it at, and run it through various levels of processing until your AI reliably spits out something you want to see.

This is hard to do, so there is a lot of monetary value in creating one, hosting it, and offering it to others. At no point do you have to create anything of value, you just have to take what other people made and extract surplus value from it. And you do this while interacting as little as possible with the art you've used for this massive dataset, or god forbid with the artists that made it.

In return these artists get nothing, usually. Not so much as a note of acknowledgement in the readme.

Human artists can and do steal from, plagiarise, and piss off other artists all the time. But they cannot do it at the scale of AI art, nor can they completely divorce the original art from its context. Even when this is done for financial gain, it's on an entirely different level, with entirely different stakes. I'm not talking about copyright law or intellectual property, I'm talking about large scale labour devaluation and deskilling for the sake of corporate capitalism which is an inherent part of most of the AI generative art tools that are used by people today.

There's a lot of cool possibilities and interesting thought-provoking questions that have become immediately obvious with AI art and it's not in any rush to go away. I think spending hours debating the artistic merit of the end product of AI generated art is the wrong thing to focus on right now. Humans also make art that is uninspired or derivative or mass-produced and it can still be artistically valuable. Look at Duchamp's fountain, literally mass produced and likely plagiarised from another artist, and one of the most valuable pieces of modern art history we have.

It's about the scale. The decontextualisation. The capital gain for a generally uninterested company seeking to extract surplus value from the unconsenting labour. The picture that gets generated at the end is kind of the least interesting part of the discussion.

#long post #ai art #gay sex cats #it is possible for an individual to create and effectively train an ai model and you can argue about the artistic merit of the result #of that process if u want. when i make art it's primarily digital and i consider the stuff i made thats supported by tech to still be art #so i dont think i have a leg to stand on to claim an individual developer who made an ai model isn't an artist u feel #its how the dataset is gathered for me. and who benefits from it. #like it does kinda boil down to respecting the labour for me idk #and as an end user of someone elses ai model you have been robbed of the ability to respect the labour that went into making the data #bc it got scraped and turned into a big blob of data thats hidden from u #whats the point in attacking the individual when u can attack the system that created the situation in the first place

[ultravioart]

Oh God damn it.

Please boycott Palworld, apparently Pocketpair (devs of Palworld) a few years ago(?) made a party game that hosted a generative ai art model. Fuck that. Those models (Stable diffusion, etc) are created by abusing workers, and use unethical openweb datasets (private medical data, NSFL material, copyrighted material etc all gets used in openweb datasets).

I instantly drop any series that flirts with ai. Why twitter.com waited until Palworld's release to post about the past ai bs is beyond me, I would have liked to know that before purchasing. :/ Palworld was announced YEARS ago.

I had honestly thought Craftopia was the first game Pocketpair released because the YouTube videos I had watched of playthroughs had said it was thier first game release???

And I thought generative ai models were not allowed on steam?!?! apparently Steam changed the rules to allow generative ai models/assets in games, but only if they are tagged as using ai. Ugh. Way to worsen literacy of what ai actually means! (Populating a forest with randomized trees in a video game uses generative ai, but not the unethical kind. Same premise can be done with buildings, etc.)

Sigh. Lesson learned. :( Be sure to dig up the game dev before purchasing. This is so disappointing, I wish I could refund Palworld.

Palworld as is at launch does not seem to use any ai generated art. Generative ai that produce 3d models from text prompts are no where near the quality of the creature models, so it seems all models and animations have been done by hand at least. But with the dev's past history I am very concerned that Pocketpair may incorporate ai or worse, NFTs, into Palworld or future games. :(

So yeah, please don't financially support Palworld. I don't trust Pocketpair.

#Palworld #Pocketpair #pokemon #craftopia #text #long text #shitty ai models need to be shut down already #generative ai models COULD have been made with noncopyright material but nooooo a bunch of idiots scrubbed thr entire internet bc #they couldn't be bothered to ethically source a dataset #not to mention all the underpaid workers who had to see ai generated NSFL content to test prompt quality. #im so tired man.

[multi-lefaiye]

tbh the openai/midjourney tumblr shit is very disheartening for me personally but i'm trying not to doompost or spiral about it

but until i'm sure that toggling it off means anything i post in the future is protected, even via reblogs, i'm not gonna be posting art here for a bit i think

#multi makes text posts #i just wanna be sure that opting out means i'm fully opted out yk? #ai isn't inherently evil but those contributing to datasets need to do so willingly #and i am not willing #as are many others

[peppermintquartz]

Hey Tumblr.

Sell this .

#fuck this ai shit #i hope y'all get poisoned datasets #tumblr #artists on tumblr

[wizard-laundry]

when i say i don't fuck with ai or others that fuck with ai, that also means those of you using ai to imitate your blorbos voices

#even if it's just for your own personal use #you're exploiting the lack of protections voice actors have and helping train a dataset that will put them out of work #this post was prompted by OpenAI rolling out a new ai voice generating tool #same shit with actors and artists and writers and fucking everything else #don't even do it for the meme guys #i know biden and trump playing minecraft is funny but for fucks sake stop engaging

[loveislandthegame]

to think I thought S7 was going to be different because the first 3 vol were different. I feel like a freaking 🤡.

Also, can you tell me about your "Eddie is Alex" theory?

you & me both, anon. clowns unite ! 😭

yes, ofc. obviously this is just a theory, not facts, but hear me out :

back in july, FB made a poll on their socials, asking players if they wanted eddie to return as a love interest in a future season. idk how they could've possibly redeemed him, but that doesn't matter because the answer ended up being hell no . of course, FB must've been shocked by that. me too. i mean, who wouldn't want to romance this ... dreamboat 😐

while we’ve never had a returning islander as an LI, we’ve had returning islanders . they’ve always been redrawn to fit the new season’s style (at least before FB decided to throw ivy & hamish as is into s7's sea of clashing art styles lol)

don’t have receipts for this, but i wouldn’t be surprised if FB already started redesigning him in anticipation of the poll result being a Yes . the majority of players didn't want him back, so anything like that would've had to be scrapped...unless they change up some things, make him a new character 👀

so basically my theory is they had a scrapped returning eddie design laying around, turned him into alex, it's more about their appearance than their personalities, but apparently eddie was gonna undergo a personality transplant anyways

returning islanders haven't been 100% faithful to their OG design, we've all unfortunately seen s5 bobby, but regardless i’ll point out the differences (that aren't caused by alex being drawn at a different angle)

their chins, eddie has a cleft chin

their lips, they’re actually pretty similar in shape, but eddie has a wider cupids bow

noses. again, similar shape, but alex's nose bridge is wider

hair colour ! clearly alex is brunette, but more on that later

not a similarity, but something i thought was interesting . alex is the only s7 guy that never had reused assets from youcef, because his body is from suresh

now the actual similarities:

face structure. jawline, ear shape, cheekbones. same facial proportions as well, like the distance between their mouth and chin

blue hooded eyes. eyebags too, it's just that alex's dont have any lineart

eyebrows. same thickness and very similar shape, same brow spacing as well . look at the glabella (distance between eyebrows) shading on both of them

their hairstyles . both have a side part, a bunch of spiky bits on the (viewer’s) left side, and hair swooping behind their head on the right side

those could all just be coincidences, but there's a mystery afoot, alex’s original eyebrow colour. it caused quite the uproar when he was revealed (it kinda grew on me, i think it's cute lol) so FB gave him a brow tint when the first volume dropped . not everyone's eyebrows and hair matches 100% irl, but in the (starting from s3) LITG universe, islanders' eyebrows correlate with their natural hair colour . we haven't had a bleached brow baddie yet, and imo, this was way too noticeable to just be a simple oversight . then i realised...his original eyebrows are the same colour as eddie's

#if alex wasn't originally eddie he's definitely based on him...or was made w/ AI that had eddie in the dataset #love island the game #litg #litg season 7 #litg s7

[snewdraws]

#skillnottalent is floating around twitter at the moment, so here is my contribution! my first drawing on the interwebs (2009) vs my ember redesign (2022) 🥺💞 art is a skill born from the labor of love also here is my little spiel about how the process of making art is special to me;

Even now this Ember design is a few months old and I can see improvements that I could make. We never stop learning and growing as artists, and I find that process incredibly rewarding. Being an artist has taught me so much about myself and connected me with a beautiful community.

In relation to AI art, and how some people think artists are "gatekeeping" art by shunning "AI art"; The real value of art never came from the final product, it comes from the journey and self-expression. The ability to visualize human thought is beautiful, regardless of skill level. AI can make pretty pictures, but it can't replicate the joy that comes from the human creation. To use AI to make "art" is to dismiss the purpose of artwork. The soul and process behind the creation are what makes art special and meaningful. Anyone can create art!

I know this is a little silly in relation to Ember fanart, but humans are silly! Let's embrace the weirdness of being here on this planet and draw funny little dragons together. Let's have fun! 💕

#ember #spyro #skillnottalent #the discussion about AI art is ignoring the ethical dilemmas of the program #i think it is not fair that the program uses artwork in the dataset without the permission of the artists #i am just referring to the concept of ai art #if i tried to talk about everything i would be writing an essay ahahha i have a lot to say on the topic

[the-tzimisce]

maybe building the ai that poisons datasets is some person's version of art, huh? are you trying to define what art is? childish of you i think. not very materialist.

#i could care less about anything that happens on a computer frankly #but between the two i think the dataset corrupting ai sounds much more interesting than image generation