So... What's Going On With AI?

I love technology and how it makes everything worse over time
GET MY MERCH: crowdmade.com/...
Notations: pastebin.com/b...
Be one of the fellas: / @kappakaijuofficial
Outro Music: / @musicofphil
Discord: discord.gg/Ty3VcQq

[@violet7773]

One telling use case I heard about was dermatologists trying to use ai to differentiate between non cancerous skin and tumors. The ai began to classify any photo that had a ruler in it as "cancer", because that's how doctors photograph possible tumors and ai fundamentally does not **understand** what it is being asked

[@Sopsy_Hallow]

idk if its the same case but morphed through retelling, but i heard of an attempt where doctors gave it a bunch of xrays and tried to make a model to find out if an xray had a specific thing (maybe cancer aswell i dont remember) and while it was pretty succesful, the actual pattern it found was just looking at how old the xray is and if its an older one it had the thing and if its newer it doesnt, which makes it completely useless for what it was meant to do

[@ooofyikes7253]

Doesn't understand, *yet*

[@irregularstuff5290]

​@@ooofyikes7253will you tell it?

[@toxic_shr00m]

​@@ooofyikes7253The "yet" crew is so funny like what do you expect it to do, learn the difference? It's not being trained to detect cancer it's being trained to detect rulers lmfao.

[@ooofyikes7253]

@@toxic_shr00m yeah what I mean though is by the evolution of the system is how it'll improve, it's the same with any AI, the technology is still developing. I think you're being a bit disingenuous here

[@kaialexander6806]

One of the most horrifying use of AI I've come across are "mushroom foraging AI guide apps" where you take a photo of a mushroom, feed it into the app and the app tells you what mushroom it is and if it's safe to eat. It's basically a spruced up version of a Google Lens search. Also, it will misidentify Amanita mushrooms for White Button mushrooms. Both are technically edible, only the former has a fun side effect of killing you. And by the time you notice the symptoms, you are going to die. Painfully.

[@Jane-oz7pp]

Weirdly uneducated to say Amanita are lethal. Some Amanita species are toxic, however, white button mushrooms are also Amanitas.

[@gokuxsephiroth4505]

Terry Pratchett said all mushrooms are edible - it's just some are only edible once. AI must have been taught on this principle

[@kylegonewild]

No idea why this is horrifying and not just a potentially useful but risky piece of technology. Jim Browning (someone who knows how to forage using field guides and stuff) has an actually thoughtful video on Plant Identification apps and he doesn't outright condemn them. His primary criticism isn't even accuracy, they can be fairly comparable to a human's accuracy, it's just the low barrier of understanding about the subjects (plants in this case) that can create a false confidence.

[@matildarose]

"Don't worry, it'll get better in time."

[@matildarose]

​​​@@kylegonewild yeah, I can see the problem. AI tools on nonprofit places can be immensely helpful, but only when someone is fully aware of their limits, *especially* with fungi. The human community on places like iNaturialist is where the rubber meets the road. The problem may be in an app designed to turn a quick buck, which is what I'm assuming is happening here.

[@hazavair5755]

I'm interested in the story of the one AI chatbot that invented a customer refund policy for an airline, and how the company started fighting that in court desperately. Not understanding fundamental things - or rather, lacking the ability to perform logical reasoning - has the potential to cause immense damage to corporate interests. Remember that time that a major pharma company was impersonated on Twitter because of the dumb bluecheck changes? Now imagine how easy it would be for an AI that the company has put in charge of the pricing system to be fed a bit of bad data and immediately drop the price of everything to $0.01. For this reason, I'm expecting the surge against AI to happen the moment one of the big companies loses several million dollars because of something a techbro told them would let them fire an entire department to spend the wage expenses on another yacht.

[@dherokbattleborn]

Yeah haha. That is a huge problem with AI. The people using them don’t understand them and as such it has the potential to do unexpected things. You never really know how it’ll interpret a request. Theres even groups dedicated to taken advantage of that and figuring out new ways to trick it into doing what they want.

[@schwarzwolfram7925]

Isn't this scenario the Idiocracy scene where everyone employed at Brawndo gets laid off because they suddenly switched the irrigation and the AI overcompensated the sudden drop in demand?

[@higurashikai09]

Oh yeah, Air Canada claimed that even though the Chat Bot was installed on their website, they were not responsible for what it claimed which is bloody hilarious. They lost the court case, lmfao

[@ferociousfeind8538]

as I recall, the guy sort of bullied the AI into repeating the customer refund policy? obviously a real customer rep simply wouldn't budge in that way. I think it's fascinating that this got all the way up to courts

[@gokuxsephiroth4505]

Quick reminder to artists that Nightshade and Glazed are two programmes developed by the university of Chicago which stop it from scraping your art. Glazed prevents the scrape, but apparently Nightshade poisons the AI which tries. I'm no expert, but go check them out if you're worried about your public portfolio.

[@kylegonewild]

"but apparently Nightshade poisons the AI which tries" sounds like injecting malicious information into a system which is...a crime.

[@drmonkeys852]

@@kylegonewild It's more of making bad data for the training. Nothing malicious about making YOUR data worse for training, if you've opted out.

[@bfgfanatic1747]

@@kylegonewild It works by altering the image in ways invisible to the human eye but harmful to AI training, causing it's outputs to deteriorate. It's not like a cyberattack or something like that, it's just a measure to prevent your work from being stolen. It's less like sprinkling ricin into the AI's wine and more like growing a big yellow berry.

[@Yawyna124]

@@kylegonewild Can't enter the dataset if they don't use it without your permission!

[@Yawyna124]

@@kylegonewild Also, they're not injecting anything into the NN. The images do not inject any malicious piggybacking code nor do they contain any, they merely exist as images. The artist and UoC have no interaction with a NN possibly interrupted by Nightshade and there is nothing implicit about the existence of the art/imagery that it *must* be integrated into a neural network, nor is it being released under the guise of being training data.

[@pascalthedogishere415]

I've seen those Google AI responses make up answers that Kindergartener would

[@nessie-from-earthbound1601]

The only good use of the Google AI I've found is it summarizing information from paywalled sources (and I can't even check its accuracy)

[@iamjustkiwi]

I hate having to scroll past that every time i want to just know yhe answer to a quick question accurately. The other day i searched for "how long can painted turtles hold their breath" and the dummary confidently said 30 hours. Turns out the more accurate answer is a bout 15 to 20 minutes. Thanks, AI Edit: meant to type summary but accidentally made a better word for it

[@drmonkeys852]

The fucksmith AI reddit comment will go down in history. Ah yes, adding Elmer's glue to sauce makes it quite thick.

[@FantasmaNaranja]

google has done the best anti AI effort i've ever seen any single company or person do by showing everyone who didnt use AI just how prone it is to dogshit nonsense

[@CrazyRiverOtter]

So far I've seen it recommend serial murdering, eating rocks, eating poisonous mushrooms, putting Elmers' glue on Pizza, and smoking while pregnant.

[@toreole5831]

Also time to be real and replace usage of „hallucinate“ with „being wrong“ The machines arent hallucinating, theyre simply wrong and bad.

[@HirokuDev]

The reason why we prefer to say hallucinate is to highlight the fact that it is completely detached from reality. A child that got an answer wrong on a test is fundamentally different in how they are wrong compared to an LLM saying insane things

[@nocturneofeclipse]

@@HirokuDev idk to me "hallucinate" has a connotation of consciousness or sentience and adds an extra layer of anthropomorphism that doesn't need to be there. It's just gibberish from something that doesn't understand the meanings of words :(

[@gaiatiful]

@@HirokuDev yeah but LLM arent humans or smart beings, youre calling a puddle an ocean. machines can only be true or false, calling a mistake not a mistake is copium incarnate for techbros

[@drmonkeys852]

Actually I still like the term. It's better imo because it's not only wrong, but it's very confident in it's statements. Being overconfident and wrong is much worse than being just wrong.

[@sourgreendolly7685]

@@drmonkeys852People are confidently wrong all the time, that's not what hallucinations are.

[@capncoolio]

100% agreed right the way through; I'm a software engineer, and I feel like I'm going fucking crazy with all these AI nerds around me. It's at the point now where all I have to say to anyone is "IF IT CAN'T KNOW IT'S RIGHT, YOU'VE GOT TO DO IT ALL AGAIN ANYWAY."

[@Ninjat126]

There's no way to know if your AI-generated answers are correct, because the AI just reads a bunch of material and comes up with a lie that sounds appropriate. Sometimes that appropriate-sounding lie will match the truth, but that's just a coincidence because the AI can't study evidence or cite sources.

[@laylascrimgeour-brown1603]

On one hand, I understand the appeal of trying to work your way around programming. On the other; it doesn't matter how confident you are with your coding skills. There WILL be an error. There WILL be something that causes that error to become 50 others. AI won't fix a person's low confidence. It's a crutch, and I'm sorry the people around you are thinking the crutch will save their asses.

[@tyler-df3wy]

One of the wildest things about AI imaging is that the images it’s trained on are so overwhelmingly white that it’ll occasionally change the race of characters to seem more diverse. And since AI sucks it doesn’t really know how to do it properly, so sometimes it’ll only change the face while leaving the hands white. So it’ll just spit out Homer Simpson in blackface and call it a day

[@CrazyRiverOtter]

My favorite thing is that AI's that do that will often achieve that by tacking an ethnicity to the end of a prompt. If you write "wearing a shirt that says" after it, you can get a blackface Homer Simpson in a white "Ehnicallly Diuurs" T-shirt.

[@rattle_me_bones]

Anyone remember when Google Gemini would just... generate images of POCs in Nazi Uniform? Very massive blunder, that one.

[@metapuns9004]

GenAI is accelerating Dead Internet Theory by decades. Searching for any popular character gets you a landslide of excessive sets of prompt images over personally created art. Same with the infinite amount of product listings for clothing with random images on it. Its all become soulless slop and soon human interaction will be secondary to bot interaction.

[@Le_Codex]

AI truly feels like the metaphorical hammer that makes all problems look like metaphorical nails. It doesn't matter if there's a simpler, or faster, or more ethical solution, you just have to punch it with that hammer long enough until it either works, or you've deformed it enough that it can look like it worked

[@stonykark]

The people making business decisions about AI don’t care if they work. It’s all about making the line go up, consequences be damned.

[@modelmajorpita]

Entering a prompt into an image generator is not the "new version of creating art," it's the exact same as when you send a description of the art you want to an artist you are commissioning.

[@ghastlyweather1750]

That new Google AI response confused me greatly. I got a slight burn and had to rely on an old information art panel to treat it bc the AI response kept telling me to add cool water right above an article that said to use warn, but not hot water. And also the fact that I know adding cool water ruins your nerves.

[@yphoenix5957]

I'm simultaneously both in the know of how A.I. works and not in the know to understand how to use an A.I. media generator. I like A.i. because I like technology and I think its something integral to evolving Mankind! But... I understand exactly what Kappa Librarian man is saying when it comes down to it: I'd rather just create art myself rather then try to learn any of this, I'm to lazy... I'd rather drag a pencil across paper goddamn. But if this A.i. conundrum we have and weren't ready for persist, it greatly effects art, it _kills_ art. This is inspiring video, thanks bro.

[@digitalkaiju]

Most, if not all "ai artists" don't know what a .psd file is. I have looked into it out of curiosity, but there is no real place for ai image generation in my work. Maybe if I wanted a ton of quick bad pictures for conceptualizing characters, but like, I can sketch a character in less than 5 minutes and at least I know how to draw hands and feet.

[@bsnow304]

Your cardigans always look so cozy. How do I steal your look?

[@DavidAndersonKirk]

Hey there Kappa- greetings from the states. Really enjoy your coverage of tech/entertainment, and it’s quite refreshing to hear your takes from Australia which has arguably better consumer protections.

[@kabobawsome]

LLMs are well suited to like customer service roles, because you can do things like give it a bunch of preset information in its prompt, and then further instruct it to just direct to humans if it gets a request outside of its specifically listed tasks it can help with. Now, our current non-AI customer service chatbots can do exactly that, but they are unnatural and typically clunky to work with, so it can be very useful to making customer service bots even more accessible. Unfortunately, they are not quite there yet. The way current LLMs work is that they already have a prioritized prompt in their basic work that the developer can't alter, they can only add a secondary prompt. Hallucinations actually aren't THAT big of deal for this, because you can do things like provide information from a database or search, and then tell the LLM to put words around the result, basically, but they are still a problem.

[@bronzemeteor1234]

While some use AI to generate a picture of a tree house, I use it to generate a picture of Joe Biden performing a suplex on Thanos from hit game, Fortnite.

[@LillyxTopaz]

My boss was a huge AI advocate. Wanted us to research it, implement it into our work, talk about it at meetings. Then we got this intern. Three years into their degree, who used Chat GPT to do all their coding work, then brought it to my boss and had him fix all the errors and inconsistency of the AI. Anyway, after he was let go my boss still talks about AI but has a lot more realistic viewpoints on it's use.

[@BabyCharmander]

This was a fantastic video and I agree with like 99% of this, though I will say regarding voice AI there WAS a recent case where a singer who lost his ability to sing due to illness was able to train an AI on his own voice so he could make new songs again? And I feel like that one’s like, a LEGIT good use of it. I’ve also seen people train them on their own voices for the sake of humor. But… yeah otherwise there’s a lot of absolutely horrifying ways those could be used.

[@sarahsues]

One person on instagram is slowly finding times when the ai uses bad sources, like tik tok(so false info but it’s okay it’s false because it’s funny) for their information, like if you search up who was the first person to do a back flip, John backflip pops up

[@None-Trick_Pony]

I dream of a future where AI gets so good and powerful that you can play a Pikmin game and not constantly yell at your screen. Ooooh, you mean _that_ AI.

[@TheGallantDrake]

We, collectively, don't want to face the fact that work and income don't correlate. We're afraid of the consequences, even though people have been working on solutions to the issues that arise from this sort of shift for a while now.

[@noot-noot-pingu-noot-noot]

I hate that AI has become synonymous with ML models and generative AI specifically because there's so many cool uses for it that just aren't being used (eg. automating keyframes for mouth movement in video games, something that was super tedious for animators is now handled by a ML model so the animators can actually work on what they're passionate in) I can't wait for all these big companies to lose millions though because no one wants these awful implementations

[@swampking257]

Felt a bit strange to watch this while drawing a comic haha Great video, thanks, from an artist!

[@Th3C0urtJ35t3r]

Fastest click in the west.

[@DxGamer6767]

What you say is all correct. AI is fucking stupid, but I don't think that matters to rich people. My father is an investor and he keeps hollering on about how AI is the next big thing, I keep trying to explain to him that these ai's are barely AIs and are just LLMs that can't even think like the way we do (not now at least). And that most of the examples he states of AI are actually machine learning / Neral networks (which has been around for multiple decades) but he won't listen. My fear is that these rich people will not listen and will shoe horse this in without proper thought, like firing a lot of computer scientists and software engineers prematurely for some 1/64th baked AI programmer. Honestly this fear is driving me away from doing Computer science in the future and probably doing something very hard for AI to even replicate like pure math or physics.

[@Yam-jt3vw]

god its so refreshing to hear this take on image generation

[@flampagan1994]

I really disagree with the AI art. To say that people that use it are horrible really just is wrong, I don't have the money to spend thousands of dollars on art for my DND campaign, if I didn't have AI art, I would just take it off the internet instead. To say its Immoral and should be illegal because some people steal art or make deep fakes are plain stupid. I could drive my car through a crowded street, illegal and immoral, but it would not make other people with similar cars or the company that produces the car immoral.

[@dunjunart]

you don't have an inherent right to free art specifically for your personal D&D campaign, the artists whose work has been stolen for AI generators do have an inherent right not to have their art used without their permission. it's stealing to use the generator because the generator runs on stolen content.

[@flampagan1994]

@@dunjunart It is legal, its transformative in nature.

[@dadlord689]

This thing they call "AI" has brought so much chaos already and is going to bring even more. I can't find anything it brought in order. It sucks at any level of complexity or responsibility. Unless you could afford to gamble for some math solutions. Or you need an avatar that you'll be ok without anyway.

[@tiacat11]

Two uploads in like a week??? Losing my mind here. Blessed as always

[@AemilianaRosewood]

I must've forgotten to take my meds again. He's back, twice in a month

[@bugg6597]

We’ve been using AI for the past 20 years. In small doses. A little bit of AI helping this and that. Not entire AI products, because it’s only ever been able to assist in small ways, never shown capabilities beyond that so this push has been a huge waste of money to learn what most already knew for 20 years

[@celtickelly13]

two kappa videos in less than a month????? you’re spoiling us

[@matthewchovance4484]

OMG fr like pookie bear stopppppppp

[@TheManinBlack9054]

I'm sorry but this video is a very big miss. Very little of it is spent on discussing the actual dangers or fundamental pitfalls of technology, while the majority is just the association fallacy. It's subjective and not tangible to the technology itself. I don't like the vibes of this group of users (who are, as he also admits are not the only one) is not a good fundamental argument. "I spent some time on Twitter and I didn't like the people who used it" is not a good argument. Twitter is not a good representation of people. He also mentions different people who use it their own different way, and then goes on to completely ignore them and disregard them. Why did you mention them then? To create a sort of dissonance between their acknowledgement and then your complete disregard for them in your argument that hinges on them not existing? Listen, if you dislike it because you are afraid for your job, ok, talk about THAT, but if you do it by attacking the people using it and then make it your entire argument then people on the fence are going to be unconvinced.

[@TheManinBlack9054]

I really liked his NFT video where he was a lot more impartial and more objective, I am not a fan of hate trains and always love to see a nuanced and rational look. I really wanted to like this video too, but unfortunately, this just was a lot more one-sided and a lot less argumentaly sound.

[@electrorage4158]

Obviously this video is A.I. generated

[@Nullsparta2]

​@TheManinBlack9054 did we even watch the same video? The first 2/3rds of the vid was entirely about why it doesn't work (still takes humans to fix the problems it causes) what's causing this not to work (the things it's trying to do are extremely complex and we don't have the programming yet to account for that) and why it'll never be adopted on a wider scale/what dangers there actually are (ie making people jobless and putting out shit art) He even goes into what parts could work with ai implementation. Like it seems like you literally just skipped to the last 5 minutes and only paid attention to that

[@hooting-ton5215]

Remember: It's not artists against STEM, it's artists against Business Majors

[@misbegotten3508]

And STEM. Artists are not what I'd call the scholarly types nor do they often care for it. They have consistently, like in this very video, conflate STEM and business majors as an enemy. Which is why I don't see artists as a defendable demographic. If the attack on anything is justified just so they can get cash to draw big booba, then maybe the industry is moving on without them.

[@nessie-from-earthbound1601]

@@misbegotten3508 "I don't see artists as a defendable demographic because they don't define their opponent well out of ignorance and just want to make money off of big titty" ??????? Are you hearing yourself right now

[@AEGreen]

@@misbegotten3508I’m personally against business majors and anime avatars

[@_decaysea]

@misbegotten3508 Your response is nigh incoherent and has little to do with the comment above. You seem to be under the impression that all artists are inherently unintelligent (all of the artists in my social circle are STEM majors because businesses refuse to pay them for art) and that all commissioned art is "big booba" (which makes you sound like an R18 addict). Please go outside and maybe speak to real people.

[@SolaireHighwind]

​@misbegotten3508 Where in this video did they say "science is bad" 😂 All the artists I'm personally familiar with care about science, the criticism is how it's being used, not "discovering and inventing new things bad"

[@Methena1928]

The corporate/tech world’s AI fixation really goes to show they’d rather let a computer do a job badly but for free rather than pay a human to do a job well

[@QuantumConundrum]

Hate to say it, but it's the same for everyone, even more for consumers. You should see the face of disgust my friends make anytime I recommend something that costs money, or when I told them I actually pay for youtube premium.

[@GardenGD]

Bot

[@alalvarez7301]

@@QuantumConundrum >pay for youtube premium You're kidding, you wanna give these people MONEY on top of everything they done on the last decade alone?

[@tangerineGamer]

The funny thing is that its actually very expensive to run and maintain these tihngs lmao

[@ieatvirgins]

He pays for KZbin Premium... 🤢🤢🤢

[@ChaosSolEternal]

Ethical concerns aside, I'm sick of hearing about AI. It's like NFTs but worse. AI is being shoved into every little thing and it's annoying as fuck. My job is pushing it when it makes no sense for the product we sell. Every marketing email that hits my work inbox is about AI. I switched away from Google's search engine because they just had to push their shit AI. I'm ready for every company under the sun to kindly stfu about it.

[@lid2966]

That's a whole issue in itself. Everyone is implementing it (broken or not) whether there's a need for it or not

[@kyberion]

My dad's big on "learn generative AI" to get a job or that every hobby I do it's "why not use ai" and I actually get yelled at when I say I have ethical concerns of it cause I'm not listening

[@napkin60]

It’s worth understanding though that they’re fundamentally different issues. Both are trendy, but NFTs had no actually utility other than being something to gamble on. AI could have a LOT of utility, to the point where it would undoubtedly make the word worse with how many jobs it displaces if we don’t somehow regulate it.

[@dackattac]

@@lid2966 it shocks me how i don't hear the AI bros lament this (hell, i feel like i hear them cheer it on, if anything). because this is how the tech is gonna die: overexposure of its least flattering form wedged into where it doesn't belong until it just pisses everyone off.

[@artemisiakyrell7727]

As a wise man once said, "the future is $2, you can attach 1.5v battery to a lightbulb & call it ai."

[@elixwhitetail]

In the case of the ChatGPT lawyer, what got him in trouble was not actually having ChatGPT write a brief for him that he basically turned in without proofreading but doubling down when the other side's lawyers and the court asked if the case citations ChatGPT hallucinated were real. Submitting a filing completely written around cases that never existed wasn't a good idea, but refusing to admit fault and insisting the court take the briefing seriously is where it crossed the line. The judge gave him the fair opportunity to correct his mistake like a mature adult and he instead wasted the judicial system's valuable time.

[@redbasher636]

"Valuable time"

[@jimmycarburator2012]

ChatGPT understands the legal system

[@eanfran]

@@redbasher636 It should be valuable to you if you pay taxes and value just outcomes in criminal and civil proceedings. If you like living in a world without mob rule I think the time actually is valuable and you just dislike some nebulous perception of bureaucracy.

[@redbasher636]

@@eanfran more meant how the justice system takes so long to get cases done. Example: See how long it has taken trump to be convicted.

[@cndl_lght]

@@jimmycarburator2012Got the mamamax reference

[@D0cSwiss]

20:50 you forgot the use-case of "defaming famous people by making them say and/or do things that are super illegal, just generally kinda fucked up, or really annoying", but that can definitely be filed under "company I do not intend to keep"

[@artemisiakyrell7727]

Also the use-case of making pornography of non-consenting parties

[@crestren5996]

The frustrating thing about AI with art is that, artists wouldnt mind AI as a TOOL. The team who made Spiderman ATS made a library of outlines for an AI to replicate on the model and it gets polished up by an artist along the way. It makes the process faster and gets rid of repetitive tasks for the artist to do. But AI "art" is never about using AI as a tool, its a product in these techbros eyes. They do not care about art, just the aesthetic and money

[@EvangelistOfLolth]

Actual productization is very challenging and time consuming. You'd have to hire designers, and product managers, and developers, and actually sit down and talk to artists, and work out their use cases. Then bring in data scientists and experts who can properly map those use cases to what AI or, for that matter, any other type of technique is actually good at. Then you'd need to iterate on multiple prototypes since you won't have a good product market fit initially, and it will take years to actually have an acceptable final product. No, far easier to rebrand the equivalent of a journal paper and try to sell it as quickly as possible as a finished product.

[@hedgehog3180]

Generative AI is in general pretty good when you're starting on something new and need some initial ideas or want like a background. I also bet that image generation AI could be a useful teaching tool by like assisting with difficult things like perspective or creating incomplete images to work from or helping with lighting. Shame we live in a technocracy where things are never developed in cooperation with workers but in direct antagonism with them.

[@TuriGamer]

Art is a business.

[@TuriGamer]

@Sol_eri Nobody cares about artistic value this is about money

[@elliot_rat]

i and some artist friends of mine use it that way!

[@A1phaDrag0n]

I'm convinced one of my students used AI to research the mites we found at a field site. The information was flat out incorrect, like jarringly so. Someone fed this student completely incorrect information and I know AI has a bad issue with invert morphological features used in taxonomy.

[@dherokbattleborn]

Never understood why someone would do that. That student isn’t getting any value out of your courses by cheating the assignments. Are they expecting to get a job letting the computer do their work? Seems a bit silly to me. Why even take the courses at all if they don’t care.

[@ConifuriousTree]

Entomology is such a niche interest to study but then also cheat on. Why are you there in the first place?

[@susanthefish2559]

I hate the way people use AI for art and novels. If you dont care enough to draw or write it then why should I care about it??

[@dherokbattleborn]

Yup. Using those tools just shows they’re in it for money rather than passion. At that point they’re just taking advantage of you, the consumer.

[@3750gustavo]

There are AI like NovelAI where the AI is stupid but really creative, and it’s best used as a cowriting tool where you manually craft each sentence with AI just suggesting words and sentences

[@3750gustavo]

I learned my loved for writing with AI as cowriter, where I have full control of the story, words used, pacing, etc, and the AI only helps me choose less repetitive and simple words to enrich my text, also with time I feel like I need less the help of the AI, as my vocabulary expanded immensely

[@lid2966]

I hate that it's also so often children's books.

[@sushiroll3795]

​@@3750gustavo I feel like you could get a similar experience by just using a thesaurus, though.

[@soupcake3092]

Ai bros don't seem to understand that 90% of uses for art that isn't shitposting requires a clear and consistent vision. Characters need to be in the right spot, they can't change clothes between panels or shots, they need to look in the right direction and display the right emotion. AI image generation just isn't even close to being able to do any of this consistently, and they think it's going to replace animators and comic book artists any day now.

[@zerazukin]

On top of this i have a feeling they'll do this, realize what a massive fuck up and mess its gonna be and then rehire people to "fix things* To which i hope and pray the artists, office or IT people fleece them for all their money because i cannot state how much they deserve to be robbed of every cent for what they're doing to entire communities and customers

[@strxwbxrry_420]

The only purpose ai art should have is shitposting. I would offer the idea of grabbing a reference you’re having a difficult time finding. Even then though, ai is so terrible at basic anatomy. I saw a tapestry towards around 2021, and it was a mermaid with 11-12 fingers total and spiral nipples. It ruined the entire tapestry. This shits only good for shitposting

[@TheManinBlack9054]

I mean all of that is possible with Photoshop inpainting. And there are consistent characters with LORAs. I don't think think those are insurmountable challenges. You need to dig deeper. Attack the actual point that you have a problem with because sooner or later all these other points might be overcome and then you'll have to move the goalposts.

[@roycampbell586]

​@@TheManinBlack9054 I refuse to dignify AI by giving it credit for things it *could* do but hasn't yet. No artist gets to receive acolades for skills or art they haven't made yet, so why do we have to give AI credit for things it can't do yet?

[@OrgaNik_Music]

@@strxwbxrry_420 No, it should not be used for shitposting either.

[@AmeliaMastervally]

Also worth noting a Harvard investigation found a lot, if not all, image generators were trained on child sexual abuse images. Like when you scrape images from across the Internet you're going to get shit like that mixed in but the fact that no action has been taken to correct it is sickening. Also a lot of those generators are being used to create incredibly realistic sexual images of children. Like I don't think I have to explain why that's gross as fuck right?

[@undefinederror40404]

Holy sht I had not even thought of that at all, but of course it would be used to generate content of (child) abuse that disgusting people get off on... 🤢

[@AmeliaMastervally]

@@undefinederror40404 honetsly I'm a huge fan of cute chibi style characters in anime which unfortunately comes with having to see them get sexualised a lot on forums and other places. I have seen, with my own two eyes, people generating images of extremely young children in explicitly sexual poses and situations including fully viewable genetalia. The fact that these programs allow you to type in prompts like "toddler" and "naked" together is disturbing enough, but the accuracy of these imagines? It genuinly fucking scary and disgusting, especially when you if you remember these photo realistic models are trained ON ACTUAL FUCKING PHOTOS OF CHILDREN. Obviously Ive reported any posts and accounts sharing this sort of thing, but just because I no longer see it doesn't mean it does e, ist anymore, and as these models are refined they're only going to get more realistic and popular with predators. Not to mention the ability to deep fake sexual content of real children. The exploitation of children on the Internet is bad enough with family channels who cater to predators, but with these tools a hand children have never been more at risk and there are no laws or regulations against it yet.

[@gaiatiful]

im certain the percentage of ai generation models that did this on purpose is worryingly higher than youd expect

[@_kaleido]

I’ve already seen people defending AI generated simCSEM because “it’s not actually real.” way more people have these sorts of fantasies and desires than we’re comfortable thinking about and it’s going to become so obvious in the next few years

[@bashartz]

Wait can you please link a source? I'd like to have that

[@slanax]

I love when the search engine tells me "here let us write a summary of the wikipedia article with AI so you don't have to check wikipedia yourself. btw we are fully aware that this will be wrong sometimes so you effectively still have to check wikipedia to fact check. but if you want a machine that makes up sentences to guesstimate the answer instead of getting the actual answer for some reason, here you go. the future is now" this is a duckduckgo subtweet

[@LiveLM]

Lmao so true. At least DuckDuckGo asks you to click on it before it'll start spitting out slop, unlike Google's latest updates 😵‍💫

[@tangentfox4677]

I like how you say this as if Wikipedia is accurate. :3

[@theencolony5595]

​@@tangentfox4677 that's the thing, wikipedia IS generally very accurate. But you can't just take random snippets from the page without causing context issues, which is exactly what AI does

[@Oreli]

​@@tangentfox4677 i mean for most non-pop culture topics Wikipedia is pretty reliable

[@1980woodpixie]

@@tangentfox4677 Wikipedia is actually extremely accurate. They have very high standards for editing and all vandalism gets reversed almost immediately

[@manic_ferret]

Generative AI can be tacked onto the long list of things "tech bros" have murdered in cold blood. But its so much worse this time, because instead of having someone tell me NFTs are the future every other day, now every site you go on has a corner of it where you will be up to your ears in bots and generated images. It is a plague lol

[@sushiroll3795]

If it wasn't for minus tags, I legitimately would've gone insane at this point. And not every site even has them, sadly.

[@randompromises1038]

But since NFTs have died down and become less profitable, there's hope generative AI will do the same right? Am I being too hopeful?

[@manic_ferret]

@@randompromises1038 god i hope so

[@defaulted9485]

​@@randompromises1038it is. The most honest person documenting selling AI image on Etsy lost 4-5 bucks in a month in the process. Thank God for his honesty. That means people profits off AI by denying customer service or just scam people by not sending the physical media. The AI demand is extremely low at 1 image per month despite generating it at 15 images per hour. That is the outcome for people trusting peddlers and course sellers to "make AI business". Some documentaries tracked what "business" these peddlers gets their money from and it nets to ghost KZbin channels reading TTS inputs with recycled content like Ask Reddit or fake news or unsubstantiated pseudoscience babble. You can guess the channel just by looking at the generic corporately designed profile pictures.

[@bobbiesterling574]

the AI dream : AI does stuff we dont want to do for us so we can focus on more fulfilling endeavors, like art for example the AI reality : a huge chunk of AI development is based around completing complex and valuably high level tasks, like art for example, forcing people out of work and only making it easier for large corporations to make money, and making us to the the work we dont want to do (sometimes for the AI)

[@vrubin]

As an artist, thank you for standing against Computer-Rendered Artificial Pictures (CRAP)

[@infinitysnapz]

the sooner this System-generated Hyper Intelligent Text stops flooding the internet, and becomes a thing of the past, the better honestly

[@risingmoon2456]

That's a brilliant anagram for that

[@popneppopnep4357]

Oh please the movement is cope at this point.

[@TripeDemo]

Why does anyone need to stand against that? If people like the outcome they'll use it, if it's not that great, they won't.

[@TheManinBlack9054]

​@@TripeDemopeople are against machines automating their jobs, they are protesting and outcrying against that welcome everyone to join them. The issue here is the affected of livelihood

[@eleeka8565]

The single-minded decades-long persistence of tech and corporate greed in commercialising but simultaneously devaluing creative skills exclusively in ways that eliminate space for the actual people component of that creativity would be impressively comical if it wasn’t so deeply and fundamentally Evil.

[@TheManinBlack9054]

Every place has been getting more and more automated, it's sad that when the common worker gets replaced he's told to "learn to code", but when artists are replaced it's supposed to be different and more important. That it's an outcry.

[@cobleen3982]

I do find it interesting how they are pushing AI on creative fields, when it is much more suited to work in tech fields. But I think it’s pretty obvious why they aren’t showing how easily their own jobs could be replaced…

[@gokuxsephiroth4505]

@@TheManinBlack9054 Tbh, the difference is that every cashier I've ever known would be happy to be replaced by machines if they were appropriately monetarily compensated. Every artists I've ever known would be heartbroken to be replaced by machines, regardless of if they were paid.

[@koumorichinpo4326]

@@TheManinBlack9054 it is when its nothing more than the largest scale act of plagiarism of all time

[@AleksandarBell]

@@TheManinBlack9054Homie, what are you talking about?

[@mirrorimagery]

part of whats so awful too is that ai image generation is often used to generate "art" from smaller creators that could have easily been achieved via a commission, but the people in question who use that sort of thing tend to not respect artists and dont like their prices. which is especially foul, considering a lot of artists who have commissions open genuinely want to make art their livelihood. in addition, voice generation has also been used explicitly to steal voicework from voice actors who have explicitly stated they do not want their voices used in this way. yuri lowenthal comes to mind. i believe there was also a case made by scarlett johansson against a company that trained their text-to-speech off her voice after she denied them permission, and proceeded to call her voice "not her property" after she contacted them about it. yikes. this stuff never gets used in good faith, it seems.

[@ParaFox404]

It's infuriating because that could have been it's job, a cheap hackjob offered by the artist on a basic bot, you pay signficantly less and get what you paid for, and can run it again and again like a nitwit till it looks right payin each time you go, then hand it to the artist and ask for a touch up to the ones you like to bring them up to scratch for a *cheaper* but still reasonable for the artist price, or if you're anxious to go back and forth commisioning a big idea with them from scratch, or not sure their style will suit the vibe, you run the bot and give it to them as a rough reference with marks on what you like and what you dont. if the artist has disabled commissions for whatever reason, the bot keeps working, real passive income, that's what they should be doing, but we made a plastic dystopia so they were used by theives and bandits to screw over the people that create the only things that people care about anymore, *the content* It would have made everyone happy, could even collab and mix bots with other creators to make wacky hybrid styles, the cheapskates and people who just want crap to iterate on because they struggle to start won't trace or try and swindle you, they'll get a hearty discount and stop whinging the prices are too high, (they won't but i can dream)

[@heartache5742]

art being a livable profession no longer makes sense economically, and while the music industry has allowed tech to subjugate it by agreeing to have their music streamed for pennies, the people who draw and act were too stubborn, so ai is here to outcompete their art

[@gerdost]

A while ago somebody told me the job I was studying to do (accounting) was going to be replaced by AI It hurt a little inside because I have been studying that for 6 years at that point And then i realized that If you ask any ai for foods ending with -um you end up with bananum or museum

[@muwuriel8231]

Maybe I’m an AI because it took me a while to come up with “gum” lol. My first thought was “serum”

[@BlueKanary]

@@muwuriel8231 Bacum

[@Kralisedra]

ah yes, my favorite food...fried museum

[@legateelizabeth]

Hey remember when Pink Floyd hosted an art competition to make a music video and some AI slop won? I don’t hate AI image generation. Well, I do, but that’s not why I’m upset. I’m upset because I’m watching literally everything and everyone that isn’t _an artist themselves_ accept and endorse it. Some even lie about it. I understand traditional art won’t ever go anywhere, we’ve been doing it since we were in caves. But our entire media landscape becoming that and it being gleefully endorsed by the people with power over it? That upsets me.

[@undefinederror40404]

Every time any AI ""art"" makes the news they always 'forget' to mention that it's trained on stolen work. At least all news reports I've seen, even ones that were made explicitly with the intention to explain what AI does, never mentioned the theft.

[@rynabuns]

I love how in the "how it's made" video for that Pink Floyd AI garbage the thief tried so incredibly hard to make it seem like he did much more than type a prompt; thieves love the word "curate" because so much of it is unusable garbage 😂

[@winteriris13]

ironically the opposite also happened, there was this ai image contest and a real photo won. i think contests in general will definiantly need much more moderation

[@GreiH]

I hadn't heard of that, but I did hear about the 6 AI images that got into the top 300 for this year's Pokémon card contest. They got removed because the person broke the rules (most obviously trying to avoid the 3 submission limit by submitting under at least 4 very similar names)

[@vaiyt]

The people judging that competition are fucking rubes.

[@lokmister]

I think my problem (other than the stealing from artists) is that they are using AI to try and replace jobs... people actually want to do? Why are we not using AI to get rid of the lame boring jobs that no one actually wants to do?

[@thinkingoffaclevernickname1928]

because its expensive to hire creative people that are also competent

[@Cr3zant]

Fundamentally? Because they've already tried. Mass production factories utilizing nearly exclusively machines that assemble products in the exact same way every time, self-checkout lines at grocery stores, using drones to deliver products instead of truck drivers. Those that worked out stayed, and those that fundamentally just couldn't be done by machines was phased out. The closest thing I can think of as a job everyone hates that's not replaced by machines is like, fast food workers but the thing with that kind of job is if you phase that out with machines then you are left with far more unemployed people on the streets than potential creatives or experts advancing technology.

[@TheWrathAbove]

We have been trying to do this since automation has been a thing. The more a company can successfully automate the more money they stand to gain. The problem is that's easier said then done. Self-checkouts are being rolled back because they aren't able to prevent theft, and self driving cars just don't work. That's not to say automation hasn't succeeded at all, much of the manufacturing industry has eliminated tons of labor work, same with construction. What we're refering to as AI (it's not actually AI lmao) is just another step in that proccess

[@takimi_nada]

@@Cr3zant when it comes to fast food workers being replaced with AI, is that AI likely can do the job, but the machinery necessary for it hasn't been developed, and is unlikely to be developed quickly

[@DavidJCobb]

they _are_ removing "boring jobs." it's just that to an AI bro, it's creativity, insight, and empathy that are "boring." to them, there's no higher calling than business and management, and to their thinking, throwing artists into a warehouse would make those people actually useful

[@fleggy5050]

There is this guy who was a graphic designer for some company and they just took all the stuff he made for them and fed it into an ai, and then fired the guy. That sounds so incredibly illegal but probably isn't because AI isn't accounted for in law yet so screw that guy I guess. (Then again AI isnt perfect and that decision of firing the guy is arguably very stupid because the style will advance and improve nowhere but its still incredibly shady) Video link: kzbin.info/www/bejne/i2PZomyCisd3fdU

[@haruhirogrimgar6047]

Except currently under law you can't copyright an A.I. made image/work. It is considered public domain. So if they hired him for his portfolio and are just gonna use his art internally, that is really sh'tty but legal. But if they plan to use his art for anything they want to monetize, then they just need to be reported to the proper judicial authorities.

[@Prolute]

@haruhirogrimgar6047 Being in the public domain means they can do whatever they want with it.

[@luancosta199]

That decision is really stupid and short sighted bc the image model will eventually canabalize itself when it runs out of his art and starts using its own images, as have countless other image models done before lmfao. Not only that, graphic design isnt just abt shitting out an image, it is quite literally 80% research and 20% creating. So when the model eventually starts generating completely nonsensical and outdated content, wtf are they gonna do? This will be completely obsolete in AT MOST a year and that's being way too generous. Ai is a useful TOOL, not an employee. These companies are so stupid 😭

[@takimi_nada]

@@Prolute the problem with not being able to copyright your logo is really transparent to me, not sure why you have more likes than the other person

[@sansmoraxz]

@@haruhirogrimgar6047 I think I remember him mentioning it wouldn't work out like that. At least in UK.

[@saucyboi1065]

Ah you thought you wouldn’t get my American view by posting at an unamerican time? You thought wrong!

[@Notfallkaramell]

Insomnia?

[@Sirvaria]

There's also, regrettably, a huge usecase developing in the ad world - no need to pay shutterstock (and thus a photographer, photo editor, model, equipment, and all else included in the price of that stock image) if you can just use midjourney to render a very realistic photo of a very specific image motif in less than 10minutes. It's goddamn bleak over here, ngl

[@abzurdo]

it's their best use case but also the easiest one to outlaw because there are a bunch of consumer protections around product photography already and you can easily explain to boomers writing laws how fucked up is to sell stuff with fake images

[@Sirvaria]

@@abzurdo oh sure, but what i also mean is - not product images themselves but mood ones. Imagine a flyer by a local insurance, depicted is a smiling couple on the cover, another man signing some contract or other at a cozy little kitchen table. These are what is so easy to replace, why bother with either expensive shootings or searching through an image database for ages if you can simply create these adhoc? The same goes for interiors - say you have a product and you want to show it in a living situation so it feels more approachable - the product itself you still may have to photograph or render with a program like 3dsmax/a professional product shot. But you can save money here as well if you skip rendering in an environment or taking more staged pictures than necessary. Just have a cutout studio pic of your product and generate the backgrounds, edit the product back in et voila. Honestly, working in advertising rn is truly infuriating, so many jobs are being subliminated away, and so many people would usually be paid for the tiniest of pictures. But now it's all just a button click. Of course this advancement is basically inevitable but it drives me up a wall that the theft of the very images that are used to feed the algorithms are used to replace the people who made them.

[@Rachopin77]

Im an illustrator, and ai images are actually way worse than stock photos to the point of being practically unusable. They don’t actually “know” what things are so using them for reference is basically like looking at a drawing that someone who doesn’t understand anything made by copying from something else lol.

[@sourpuss1455]

I really wish they'd focus AI on important things like identifying cancer cells, as opposed to making sloppy art.

[@arturoaguilar6002]

They tried; but AI kept misidentifying it.

[@damienwit]

That already exists, but it doesn't work because AI does not fundamentally understand what it is looking for.

[@andrewliu6592]

people do do that but that's more in research circles so most people don't care much

[@Meow_F]

I’m an illustrator and it’s hard to show work I’ve made for free on Twitter with the risk of it being scrapped and stolen, it’s really disappointing how comfortable people are with it

[@roycampbell586]

Nightshade your work. Break the machines.

[@batfurs3001]

You can glaze your work so that ai image generators can't use it

[@gokuxsephiroth4505]

Look up Nightshade and Glazed. Those programmes apparently stop it from stealing your art

[@misbegotten3508]

This has ALWAYS been the rule of the internet. AI hasn't changed that. I always have to doubt the pleas of the "downtrodden" like this because you're swearing the rain is worse when it's the same as it's ever been. If you don't want something "stolen", something impossible in digital space, then don't post it on the internet. Why are you entitled to post anything on it? The lack of introspection from many "activist" artists is what makes me indifferent if not hostile to your thoughts.

[@nessie-from-earthbound1601]

@@misbegotten3508 Have you considered that people just taking shit off of the internet, even without AI, was still a bad thing Also, the "why are you entitled to post anything online" can be flipped onto anyone for any reason. Like so: this AI stuff clearly just pisses you off, so why did you post on a video about it and just piss yourself off? Could it be....because you're entitled to post about what you want? Like everybody else is?

[@404_kayjay_not_found]

Ah yes, the classic dream of machines taking on our labour so we can pursue higher items on Maslow's pyramid of needs, yet here we are, doing meanial tasks for minimum wage while computers make art.

[@matheussanthiago9685]

I fucking hate this timeline

[@SansToast1]

Ai? More like artificial unintelligent

[@lazydelibird]

Artificial Idiot is also a tempting option

[@daydihey3045]

gottem!

[@TheManinBlack9054]

I think its very preemptive and even petty to discard and discredit the AI so fast and so early. Calling it unintelligent and other words for not smart people as the person above did (I can't use the actual word or YT will delete my comment). We are still early, it's development has only started and it's already giving very promising results and big leaps. You are laughing at a baby who is only starting to walk for not being stable enough. You should extrapolate and think of the future and not think that all it's going to be is this. Already AI can do so much and be useful for so many things, being so dismissive of it and its potential seems very near-sighted to me.

[@ica5115]

GOTTEM

[@brorthor6508]

@@TheManinBlack9054I think the folks above you may have been joking

[@villageidiot7584]

"Once men turned their thinking over to machines in the hope that this would set them free. But that only permitted other men with machines to enslave them." For how popular the Dune movies are, it seems some of the main themes Frank Herbert included in the book have been lost.

[@murisbukvic2496]

One reason why AI is generally a failure in assisting with google searching, generating "art" (and just doing any kind of menial labor in general) , is because it can't think of a solution to an any given problem that leads to an improvement, or a solution that doesn't involve breaking some rule. AI can't discern that stealing is bad, that someone might object to the solutions that are either fradulent or non-existent, or even that their solution might hurt people. It takes orders from their master, whoever that might be, and applies the solution in the most literal manner imaginable, unless limitations are set beforehand (and even that can go so far). It's like telling a kid who never went outside in his life and doesn't know anything about how the world works to go to the store and get a loaf of bread. The kid ends up stealing it, because while to anyone else, "getting it" means "buying it", but that kid doesn't know the distinction, and the next thing you know, there are cops at yoursl door. While the kid can be taught the concept of theft and laws and context of both, an AI can't do that, because they have no idea that fundamentally, breaking a law to complete a task is a worse outcome than doing it legally. They have no concept of legality or improvement. And teaching an AI to work around the legal frame, in context of betterment, in a way that doesn't take forever to do or breaking down is so complicated and time consuming that it's just better to hire human labor, because people understand context in a way AI never could or never will. But hiring people costs money, so obviously companies are going full throttle with AI. Quality Assurance? what is that?

[@TheManinBlack9054]

I don't think you understand how AI even works. It doesn't take orders from anyone. What do you think it is????

[@roycampbell586]

​@@TheManinBlack9054What would you call a prompt other than orders? You could be cute and call it a sees or something but the utility is the same

[@raghnall847]

​@@roycampbell586 Ignore him. He's just a troll

[@drmonkeys852]

​@@TheManinBlack9054 He does get at a fundamental problem with AI called "alignment".

[@sourgreendolly7685]

They do pay people to test AI, they just also want to cash in on the trend now and are pushing it out far more than it's ready for. It's one of the countless ways the development of AI is being mishandled. It's always profits over ethics with corporations. They have shareholders that don't understand technology and they want the next big thing at all times.

[@SirGrimly]

As I have always said with AI image generation: The people most capable of bringing out the most of the program are the people who want to stay as far away from the program as possible... prompts are only as good as those who write them lol

[@thop7964]

I got a M.S. recently, my research was on machine learning. Pretty much all research in this area including my own, has a conlcusion that goes "Yeah we've tackled an issue that hasn't been approached this way before, and we've improved upon the previous results of non-ML models, but for the love of god we are still so far away from actually having a high enough success rate please improve it further before using it good god." ML is a subset of AI, so saying what we're looking at these days is AI is technically correct, but I feel like the general public hears "AI" and thinks of a sentient sci-fi robot instead of what is essentially a big equation.

[@mariokarter13]

AI is the new NFT. Even if the core technology is solid, it's being run into the ground like a fad.

[@drmonkeys852]

Though unlike crypto and NFTs there still is some usecases with it, so major corporations actually care which makes it worse.

[@mkeller5405]

​@@drmonkeys852While I understand what you're trying to say, crypto and NFTs do actually have their usecases to the comparison is incredibly apt. For example NFTs can be used to help online artists sell their work through proving who the artist is as well as making it "unique." And crypto has been used online for decades and is near impossible to countrfit.

[@drmonkeys852]

@mkeller5405 no it's not even close to the same level. AI can help you code, write documentation, draft ideas, and summerize text. Crypto was trying to replace things that we already had working fine in a worse way that enabled scammers and con artists to thrive. Artists already sold their work without issues, and surpise never used NFTs because they don't need to. The only remotely positive use case is for foreign currency exchanges, but even that is very niche.

[@userwasneverfound]

​@@drmonkeys852this.

[@1980woodpixie]

@@drmonkeys852 "Crypto was trying to replace things that we already had working fine in a worse way that enabled scammers and con artists to thrive" Replace the word "Crypto" with "AI", and replace the word "Was" with "Is" in that sentence, and it`s 100% factual.

[@atmamaonline]

Technology must always be evaluated by how easy it is to commit fraud with it (in addition to other things (none of which are ever considered))

[@JensDoll]

Guns? Anyone?

[@TheRealPentigan]

There's a lot of crimes I can think of you could do with a gun, but it's hard to define most of those as 'fraud'. I guess insurance fraud?

[@TI3RU]

there's a type of "ethical AI" that has existed in Photoshop for years now, mostly through the patch/Band-Aid tool and the content aware fill tool. Both of them just use the art that you already made to fill in little spots that you need, but that doesn't have the power to steal entire works from artists so no one talks about it because all the AI "artists" want to do is steal from artists😂😭

[@Nassifeh]

There's so many *little* things these tools will probably eventually be useful for. Like I think there's a great use case for using it for animation lip flaps. Little time-consuming things that are repetitive but not identical enough to be done with previous tech! But then you look at that stuff and you go--oh, um, these companies are being valued at BILLIONS of dollars. Nobody's spending BILLIONS of dollars on these things. So where's that money eventually going to come from? That's when I go, oh no.

[@dackattac]

i read about a guitar amp modeling plug-in that used machine-learning on how various mechanical/electric components influence sounds to get more authentic-sounding virtual amps. these things definitely have cool use cases! but there's some very morally ghastly ones! i'm still angry lalalai, (which you can feed MP3s into and it spits out isolated vocals, drums, bass, etc) just nonchalantly sent out an email blast like "hey, now we also offer audio deepfaking"

[@henryfleischer404]

Another one is the precursor to most image generation AI, de-noising AI. It's good for reducing render times in Blender while maintaining good quality, so an image with 4096 samples per pixel can look slightly better.

[@thisnightsrevels9605]

It's also horrific for the environment! I feel like everyone glosses over that. A single chatgpt prompt is equivalent to buying a 16 ounce bottle of water, pouring it out in the dirt and then chucking the bottle in a bush

[@iamjustkiwi]

Tbf that was already how I started all my research, but I can see how it is a problem

[@jazzwilliams7040]

He says on KZbin, which uses massive amounts of resources continuously in order to maintain a non-profitable amount of videos and stream them to viewers across the world. 🤔

[@sushiroll3795]

​@@jazzwilliams7040 "We should improve society somewhat." "Yet you participate in society! Curious... I am very intelligent."

[@jazzwilliams7040]

@@sushiroll3795 The point is more that because “AI is bad, me no like AI” mentality the person looks around for other reasons to support why AI shouldn’t be a thing, and lands upon those that they would not apply to things they do like. Which is quite disingenuous. And who are *you* to decide that AI is having a more detrimental effect on society than YT, or any other myriad of technologies that use tons of resources?

[@kylegonewild]

@@sushiroll3795 "Without any evidence I'm *claiming* this would improve society somewhat" "Source?"

[@amymcdonnell5084]

no way librarian jesus watches basketball now

[@Vennyaki]

Being an artist and a computer science major/programmer is so funny. Techbros are stomping all over the compsci portion and showing that they don’t understand their own technology at every given chance, all the while screaming “this is the future” (for the record: generative AI is a hype bubble. It’s NOT long for this world and IS going to burst). Then your other interest (art) is ruined because techbros (see above) are showing how uncreative and lazy they are in every aspect. The complete and total disrespect for the arts is the point. They hate artists and want to make it our problem. There’s two wolves inside me and both want to scream 🫠

[@henryfleischer404]

Same here

[@duck9271]

I know I’m late here but could you explain to me why you say generative AI is a hype bubble? This isn’t me trying to oppose you, it is a genuine question. I’m actually trying to get into digital art, and I’d love to have some hope that my skills won’t be useless in the future like everyone keeps telling me.

[@Vennyaki]

@@duck9271 I had a fully worded mini-essay just to express my distaste for generative AI, but I literally can't post it because KZbin deletes it within 2 minutes. I don't know what words I said that would be flagged. So instead, I'll have to ultra summarize my points and ask you to look up every bullet point I write. Really sorry about that, I really would've preferred to just have the essay here. Basically: - Too many similarities to the 2000's internet bubble - OpenAI is running out of content - Model collapse (What happens when the above happens, meaning AI now is as good as it gets) - NVIDIA chip prices - AI winter - AI lawsuits (legislation could be an issue for AI) - Rodney Brooks (an AI and robotics professor who knows how these things play out) - Too many startups advertising AI, only to be using OpenAI models (this one may not yield results, but in the tech scene, there's so many job openings and comments regarding this. It's oversaturating the market) I believe any number of these things could easily make the bubble burst and/or predict it happening. Feel free to ask for clarification on any of these. Maybe I can elaborate in smaller comments, if needed.

[@duck9271]

@@Vennyaki Thanks for elaborating! These are some really interesting points. The one about model collapse is especially interesting to me, considering the statement “it’s only going to get better” felt like one of the main arguments for AI. I’ll definitely be doing some more research now. Too bad you couldn’t paste that essay, I absolutely would have read it lol.

[@Vennyaki]

​@@duck9271 Yeah, no problem! The essay I had was pretty in depth, so I guess it scared KZbin a too much or something lol. But yeah, that whole "it's only getting better" argument is proof that techbros either a.) don't understand how generative AI works, b.) don't care and are closing their ears in the hopes it goes away, or c.) both. Generative AI can only be as good as its content. OpenAI has taken the majority of the internet and is actively running out of things to use. A growing amount of internet content is nothing but AI now. Know what happens when you mix every color in the rainbow? You get brown. AI images are already a mashed together soup of everything it has ever found. What we're going to start seeing if this keeps up is a photocopy of a photocopy effect. At some point, AI images are just going to be completely unintelligible smears and no amount of stealing is going to fix that because it'll be so ingrained into the model. The only option would be to completely redo the model but like... We'll just wind up back to where we started tbh? So that would be fruitless. What we have now IS the best model, it can only get worse from here.

[@laudatesolem2185]

As an artist, I don't understand AI bros telling me shit about "we need to make sacrifices to push progress" All I see is people defending the right to put creative expression on the conveyor belt, while someone else handles the expenses of that choice. Surprisingly many of fierce AI enthusiasts don't want AI to assist, they want a straight up replacement of human resources. They just want to consume tons of artistic content because they think they are entitled to it (like... hello? Maybe we’ll cover our basic needs first, and then we’ll profit from something that doesn’t necessarily need to be fully automated?) AI is a good tool, but people, as usual, use things they are given in the worst possible ways. Many of those who work with generating images just want to enter a word and get a ready-made result. This is terrible laziness and a lack of desire to create. As a result, they don't understand how to create and what it takes to convey a thought into any medium. (And yes, for some reason, of all the professions that need AI help, people have chosen to HEAVILY profit directly from artists. And I say it is to profit from them, not to help the creative professions. The use of AI in many ways is associated with malicious intentions tied heavily to pure laziness or greed. From companies especially. These intentions rn are not justifiable for me personally) Sry for ranting and bad eng

[@limpwrist3592]

AI has its uses, specifically in accessibility. Be My Eyes and Seeing AI are FREE apps, that help blind and low vision people get a glimpse of their surroundings. AI for profit is gross, I don’t think that should change.

[@Huweeey]

@Depressionwave2338 Too bad that kind of tech nowadays doesn't really appear on mainstream media, most of the time it'd be rather convenient products or literally things and features we've never asked for. And 60% of them are probably there for money laundering.

[@Nassifeh]

Even there, I'm a little worried about it in the short-term, because I think the error rate is just way higher than I think we'd tolerate for accessibility stuff in other respects? I can't imagine going through the world getting told what's happening around you and knowing that there's a pretty high chance it's wrong. Hopefully that much will improve relatively quickly, but we do need to be careful about giving people accessibility tools that WE think provide an experience they actually don't.

[@TurnipTheBee]

as an author, thank you for not encouraging this generated shit also 1) if you're using chatgpt to write, it's going to be shit. it's not a good author, and it doesnt know what to do 2) if you're not, please take a careful look at what youre reading- chatgpt has a very.... """unique""" writing style, and we need to catch people out with it 3) if you write with it but dont claim it as your own, more power to you

[@penguinthelad9566]

Holy shit, 2 uploads in a month? Christmas has come early in the form of an Australian librarian

[@princekyle4132]

Current “AI” is just terrible autocorrect. And often times, autocorrect can’t even work right! What’s scary is the thought that its stupidity will take over, not its intelligence like in movies :/

[@NCSGeek]

I can't say this without at least slightly sounding like an annoying know-it-all but a big part of this is that it's being oversold and most don't understand what it actually does. I'm not saying I understand it 100% either but I do have some computer background and I've looked into enough to come to a decent understanding. It's not intelligence. Computer learning algorithms have been around for a while now but it took GPT, a very large model, for it to come to the mainstream. Now all of the out of touch companies are pandering to investors, neither of which understand enough of it to meaningly use it or prevent it causing more harm than good.

[@AGNAB712]

the analogy that ai is like lead is an incredible way to put it and makes it a lot more digestible

[@dackattac]

please do not digest lead

[@arturoaguilar6002]

10 to 1 that if you ask AI to tell you if lead is digestible, it will say yes.

[@rattle_me_bones]

Unlike lead!

[@JohnGardnerAlhadis]

"AI" = Appeasing Investors

[@matheussanthiago9685]

it isn't great it has far more downsides than good uses it emits more greenhouse gases than the the Republic of Chile but hey have you consider line goes up?

[@CthulhusBFF2]

The corporate/tech world’s AI fixation really goes to show they’d rather let a computer do a job badly but for free rather than pay a human to do a job well

[@supergub1]

There does exist exactly ONE morally positive use case for the voice-alike machines, and it's those videos where the past 5 US presidents are playing video games together.

[@bashartz]

Honestly yeah, comedic parodies (when done right) can actually be considered ethical use. personally I found those videos wholesome asf lmao

[@dackattac]

or tier-ranking weezer albums.

[@TheAngryXenite]

Listening to Trump, Biden, and Obama debate mass effect endings is still a highlight of the technology for me.

[@henryfleischer404]

@@sax0catAlready exists without generative AI

[@dazeslays]

don't forget eminem's first day as a second century warlord

[@Radioactivesparks]

BABY WAKE UP, 2 NEW KAPPA KAIJU VIDEOS JUST DROPPED WITHIN A MONTH OF EACH OTHER

[@possitivelyinsane]

I heard doctors on TikTok talking about asking chatgpt for diagnosis info and so on, it's terrifying

[@ayjay2287]

10:42 I think the most telling thing about this specific example is that free online grammar and spell checkers have been a thing for years now

[@userwasneverfound]

Yes, and now they use advance MLs now(AI to you folks)

[@matheussanthiago9685]

@@userwasneverfound which somehow makes them actually less reliable while simultaneously emitting the same amount of greenhouse gases as a small nation isn't the future neat and cool and neat?

[@StCrimson667]

The ultimate irony for me is that, literally the first couple of days after the first AI photo-generators were released, I would actually say that, overall, most artists I saw were actually pretty onboard with them because they looked at these generators and thought that this could help streamline their process. Artists were talking about how these could be used to generate references so they didn't need to spend potentially hours looking for the right picture to work from or a LOT of artists were saying that they could finally add backgrounds to their works without having to struggle with them. SO MANY artists were legitimately interested in using these and were even considering paying subscriptions for these services RIGHT up until it became clear that the point was to do their job for them and then the support crumbled almost immediately. A happy medium was possible for a few moments, but the AI/crytobros just HAD to completely and utterly overlook it.

[@Giruga905]

So true, my professors in art school still see those potential uses of ai as a tool to help streamline your own work process, but literally almost every person that isn't an artist sees it's potential as replacing jobs and not having to bother with actual artists to draw something for them if they can't bother to learn the skill themselves. I partially understand where those people come from, but how can you not empathise that people in creative professions, the ones i at least remember were the ones being prioritized to not be replaced by computers, see this as rightfully a threat to their career and livelihood

[@Vennyaki]

Back when Artbreeder was the only big free AI image generator, it was mostly artists using it to make their OCs real. The real joke is that artists would’ve been fine with AI if it was made alongside them instead of being deliberately made and used to harm them 😭

[@TheManinBlack9054]

You can still generate references with them and use them for backgrounds. Literally nothing of substance has changed.

[@waffleten9750]

You know, outside of the fact that using them harms actual artist. ​@TheManinBlack9054

[@TheManinBlack9054]

​@@waffleten9750how does it necessarily harm them by you using it? What is the difference between these old ones and new ones? The only difference provided here was "now the annoying people on twitter are talking about it". If you are that swayed by some twitter people doing something that you don't like that you transfer it to the product then idk what to say.

[@MarlKitsune]

The problem with AI is not that it can cause harm, its that it was created in a time where evil isn't punished but lauded. its going to suffer the fate of TNT.

[@amoureux6502]

Comparing AI to lead deposits was actually a really genius analogy!!

[@moistenedwall1003]

Fr

[@ixeliema]

Yo, two Kappa uploads this fiscal quarter? He's gonna be rich! 🤑 (tis a jape, none of us will ever own a home)

[@xyrissavage4983]

"'tis a jape, none of us will ever own a home" is a brilliant sentence.

[@SlapstickGenius23]

I don’t think many people in my generation will own a home in any minority world country.

[@abreathingcoffin8089]

As a professional artist who has had others lecture them how "I need to adapt my workflows to new 'tools' like ai." the description of me needing to slowly cut-off every possible escape route in order to get something I sorta wanted is so accurate. If AI was a real human assistant I would not hire them ever because they can't understand the work we're doing, they keep giving me things that don't fit together cohesively and they just 'don't get it.'

[@yellowcorgi9053]

Two videos in less than a month, he's putting his whole kaijussy into these vids for us 🙏

[@lunadetta]

god, as an artist, it is just gut wrenching to see the shit people do with "ai" when they could spare like 50 dollars (cause artists need to lowball in order to find work) on a better, more comprehensive, more personalized graphic its just..... i dont like to think about how dystopian it all is LOL

[@Mrinsecure]

Part of the problem with "AI" is that it's a marketing buzzword, not a technical term. Depending on the exact context, AI can mean "a thing our product was doing for years but we're going to call it AI so we can charge you more for it" to "an overly complicated, resource intensive system that doesn't do half the things its owners say it can do" to "actually an outsourcing company in a trenchcoat."

[@DTStheTDS]

Most AI will probably tell you Morbius was a successful film and believe "It's Morbin time" was actually in the movie....I'm ok with that.

[@OmarFW]

I once had an AI art advocate argue that artists are trying to gatekeep their mediums by requiring human input and the prerequisite of building talent over many years. That's not true, but that take made me want to immediately start gatekeeping AI idiots out of the art industry forever. They're only in it for profit and they will destroy art in the name of pursuing that profit just like everything else they touch.

[@TheBiggestGame]

i was going to comment anyway but the post about the gbf artstyle took me out. as someone who (unfortunately) plays gbf, not only are those just cheap knock offs of several characters, but that's not even the correct art style. these guys really have no standards whatsoever in terms of artistic intent.

[@felixdraconic]

I know a furry artist that doesn’t think AI generating images is considered art theft T-T “what if I want art that looks like a mash up of 5 different artists styles” idfk learn to draw it instead of generating bs, “what if I want art but can’t afford it” “you don’t like AI? So fuck poor people then right?” An actual thing he said to me. Like what the fuck bro. Dude could just commission a smaller artist instead of generating shit trained on expensive artists. but he also doesn’t consider tracing as theft either so- he is the worst, unethical, least caring for the art, artist I know off.

[@TheGeekRex]

I think I despise pro-AI artists more than AI bros. Like they don't see that what they're supporting is openly coming for them, and they're already an artist so they don't even have any kind of excuse.

[@jax2903]

Sorry what? Even on a personal level, if you’re broke and want to gen something for personal use, NOBODY is “losing income” because they wouldn’t have gotten a commission anyway. And the idea of “just learn to draw it” is also so bizarre because it’s an inherently ableist take. I’m fully on board with the idea of it being illegal to monetize ai generated images using styles, but the idea that it’s “unethical” is just posturing. Brownie points on the internet don’t win you shit.

[@felixdraconic]

​@@jax2903 you clearly dont understand what its like to be an artist. firstly the comment wasnt about people losing income but since you brought it up people are absolutely losing income because there are plenty of artists who charge under $20 for unique art thats closer to what youd want than AI can get, sometimes youll even see people offering 'pay what you want' commissions where some will accept as low as a dollar or even a few cents if someone is really broke. i dont care about dumb "brownie points" think about how its effecting actual people. the point of the comment was its unethical to use artists hard work to train AI *without their permission* AND there are artists who will do amazing art for very cheap. youre literally doing the "so fuck poor people right" 'argument'. also its not ablest, to say to learn to draw it. its ablest to assume disable people cant learn to draw. if you made your lively hood on art and someone took that and starting making shitty dupes anyone could make you would get less commissions there for less income and im sure youd be as upset as artists are because it is basically theft and unethical af. put yourself in someone elses shoes for once.

[@genericyoutubecommenter589]

world's biggest cheerleader for the Leopards Eating People's Faces Party

[@felixdraconic]

@@genericyoutubecommenter589dude stfu that comment doesnt even make sense

[@lupuszero9879]

THANK YOU for calling it AI generated images and not art. I think making that distinction in our heads is the first step against it.

[@jimmyseaver3647]

It is undeniably an image, a bunch of pixels arranged to look like _something._ However, calling it "art" is, at best, extremely questionable. It doesn't matter how good or detailed one's prompt is. The moment one hits the "generate image" button, it's a roll of the dice to see if it churns out something that's "good enough". The problem is, most of these techbros have lower standards for that to start with.

[@1980woodpixie]

@@jimmyseaver3647 The definition of "Art": /ärt/ noun 1. the expression or application of human creative skill and imagination, typically in a visual form such as painting or sculpture, producing works to be appreciated primarily for their beauty or emotional power. "the art of the Renaissance"

[@Ash-bn8xy]

One of the things that REALLY sucks that you didn’t mention was people who “commission” an artist, steal the sample sketch the artist made, and use AI to complete it, so they don’t have to pay the artist 🙃

[@psyberdelicxp6042]

A.i. image generation was an interesting little thing when it first popped up. CHAT GPT showed me its limits within the first hour of messing with it. Its very telling that NFT/BLOCKCHAIN enthusiasts ummediatley pivoted to A.i. Thats all i need to know about it. Also the trend I've noticed : people most enamored with A.i. images arent themselves artists. For Practicing artists, A.i. images have very distinct "tells" and im not even talking about faces and hands. Im not saying A.i. image generation doesnt have a place. Once the dust settles, it may be useful for brainstorming, rapid prototyping, and ive seen some undeniably cool stuff that where A.i. was used. Once all the dust settles from the hype, and money men i think we will see how useful it really is.

[@henryfleischer404]

Yeah, I'm working with some people on a game, and our first bits of concept art are AI generated- this is the stuff given to the actual artists like me, who are currently working on more concept art. Pretty much like how when I commissioned some music, I found some other songs I liked to show the musician.

[@MagegirlNino]

There are so many AI images of "products" on social media websites, and gullible people saying "Where can I buy this?" Then the scammers post their malicious links.

[@ishavedoffallmyhair]

I saw a gorilla couch yesterday on amazon that looked soooooo clearly AI generated and people were falling for it. I’m amazed it hasn’t been removed from the platform yet.