Generative AI is a Parasitic Cancer

[@acegikmo]

FAQ/clarification thread: Q: Aren't some of these more likely content mill/cheap labor used to write corporate slop, rather than AI? A: Some might very well be, but, 1. the articles are polluting just as much regardless, and 2. AI will exponentially grow and encourage this kind of slop regardless, and 3. Corporate humans coercing other cheap labor humans into writing slop is arguably even worse than just using gpt directly. maybe it's just me but I don't exactly find solace in the idea of using pseudo-slave-labor to write AI-grade digital pollution. I should've probably done more research on this, but also, this was recorded in an improvised 4h ramble the day after, and I've been explicitly trying to be less perfectionist, because it leads to me never releasing anything. the downside is of course potential mistakes like this, where I should've read up on something beforehand. But I felt like I had to act quickly while I was still in it, mentally, otherwise this would be another video on the pile that never made it out. It doesn't of course change my main points, even if literally all of them were content mill stuff, but, in the end, I guess in that case I accidentally showcased a real instance of how chemotherapy accidentally targets healthy cells too Q: Is the description of the video AI generated? A: no lol but I'm gonna take that as a compliment in this one rare instance! Q: Is the thumbnail AI generated? A: I stole the green tentacles from an already generated image from stable diffusion's website, otherwise no. I figured this is the one instance I'm gonna use AI imagery because it's literally supposed to represent AI! The rest I cobbled together in photoshop as usual, using some screenshots of search bars etc. Q: How many cats do you have? A: Three! Salad is the white fluffball napping throughout the entire footage. Thor is the tuxedo cat walking across my desk near the end of the video. Finally, Toast is the guy who unplugged my monitor in the added footage! Also toast and salad are brothers. I figured this would be important information

[@dariuslegacy3406]

It'll only gets worse as these garbage AI's are trained on corporate slop and other garbage AI search results. The AI slop will only get worse and worse as the whole 'Garbage in, garbage out' rule takes hold as more and more internet content is AI generated in the first place. Salad, Toast and Thor are cute names for kitties.

[@powerfulghost]

Thanks for clarifying. I've worked in content-writing for businesses for a long time, and the quality issues you've identified predate AI by a long shot. Sadly, most people simply don't know how to write. In fact, as bad as AI is at communicating, it generally won't make the kinds of grammatical and structural errors that you've spotted.

[@valet_noir]

Indeed truly important information 🤣 love to the kitties!!! 🥰

[@mike-yii]

You don't address the point. You claim it to be GenAI, while it's not, LLMs have no problem writing grammatically correct sentences. There were always trashy websites when you googled obscure file formats. Ironically the articles could be much more coherent and informative if generated with current LLMs. Now let's see if the author has integrity and will remove false claims from the video... or join the ranks of SEO spam in quality of information.

[@swojnowski453]

perhaps you should have started your search from Wikipedia. Life is an antenna directed to the sun. If you struggle to find meaning is life, disconnect, go to a park, look at ducks in the pond. repeat for a week, then see if you are missing what you have experienced over the previous week there. The sun and the sunny weather gives meaning to life. Once you reach certain age, the only thing you wait for is another sunny day. Not sure of that? Write it down and come back to it in 10 or 15 years. That will the most important thing you will always look forward to then ;). The network can take your youth from you, but your nature will prevail, you will go back to where you belong. The sun is life, it is religion to all of us. Do not attach your life hopes to anything technical or human made, unless it is your physical creation. Tech will always will get developed faster than you can comprehend, you will get automated out of your profession sooner or later. Focus of the physical world, and place the network can't reach. The more secluded and disconnected place the better for maintaining hope and keeping depression at bay. Forget about tech, it is waste of time. Focus on growing food, painting or walks in the park, these are things nobody can take from you, and the added benefit is, you can disconnect from all the bs corpos try to stuff down our eyes and ears.

[@acegikmo]

I'm guessing this will be one of those comment sections that will quickly turn toxic so I'm just enjoying my time here while it's nice and calm. hope you're doing well, whoever is reading this

[@eugenej.5584]

What cancer isn't parasitic?

[@SueDoeNym-b4d]

you too 🐈‍⬛

[@loleq2137]

You too!

[@QuickNETTech]

Same to you!

[@Blitztein_beta]

the calm before the storm

[@arcywastooshort]

The amount of times I've had to add "reddit" or "before:2022" in my searches to get to actual humans has been absurd

[@1495978707]

Unfortunately reddit requires knowing who you are to view it anymore because they want that higher API revenue... *from these very same AI assholes*

[@fulcanelly]

why 2020 ? isn't generated content appeared in 2022 ?

[@xenontesla122]

You can pretty safely bump that up to 2022. Imagine how much worse the pandemic would have been if we also had to contend with AI misinformation at the start of it…

[@LewisDecodesAI]

​@@xenontesla122we did, that was without AI.

[@arcywastooshort]

​@@fulcanelly pandemic and ai shit might've fried my brain irreparably and time has lost all meaning to me, I forgot ai boom started a bit later

[@shaunswett6684]

In 2005 you extracted value from the Internet. In 2025, the Internet extracts value from you.

[@Druskqq]

Google has been extracting value from us since god knows when. Using google services you’re literally the product - nothing comes for free.

[@Darq_IsLive]

s/2005/1995/

[@modalmixture]

These webpages read like my most mediocre undergrad trying to sound smart and fill a page quota.

[@smort123]

"This is a up and coming file format that more people use every year" is how I tried to justify the relevance of my research papers in the introduction at university

[@seowebua]

because it is (and always was) like that on those types of e-platforms websites. All that content exists only to get users to buy their platform subscription

[@Hextrill]

That's both where AI got it's data, AND the resulting material used by the next generation of lazy students. we've come full circle on bullshitting in papers.

[@Glass_Ninja]

It's using training data to come up with, what is in many respects, the most average language possible. Mediocrity is in essence a goal.

[@SpeedOfTheEarth]

Exactly

[@tb_eest]

A browser extension not unlike uBO sounds like it might be neat. Where users can report websites that post AI garbage so they may be avoided by being highlighted or removed from search results. But if such a thing were to happen, it would also be the ideal resource for OpenAI and other such leeches to scrape even more websites.

[@LiquidEights]

it exists, there's one made by Iorate that does exactly this when paired with a custom list by NotaInutilis. can't mention it by name because this website makes my comments disappear when i do that

[@Vamaka12]

just thought of the exact same thing

[@Nikola_M]

That would be more like the extension Sponsorblock, but good idea

[@janniboy2766]

What I find frightening about the way algorithms work today is that when you get videos and news tailored to your interests, you start to think that important issues like the sketchy circumstances under which AI is being developed and deployed are already being debated and in the public consciousness, when in fact most people are neither informed nor concerned about how these technologies will affect their lives.

[@sage5296]

News echo chambers fueled by engagement maximizing algorithms has already been an issue for a while yep

[@janniboy2766]

@sage5296 what irritates me is that there are still a ton of studies that say echo chambers and information bubbles are not that big a deal.

[@cookie_plant8992]

The internet is a part of reality. How do you know that people are not all in your "bubble"? Probably due to family, friends and coworkers. Where do you draw the lines for your bubble? I think its kinda nice that not everyone is forced into the exact same perspective/bubble. Being tolerant/mindful of other bubbles is kind of separate from that, right?

[@janniboy2766]

@cookie_plant8992 i didn't say that the internet is seperate from reality. It's just uniquely structured. And the curation of media today is mainly controlled by big cooperations through the aid of complex algorithms users usually don't fully understand. That is, in my opinion, very different from the way we control the information we're confronted with and the people we surround ourselves with in real life. If that makes sense

[@pavelperina7629]

No, youtube videos are not tailored by my interest, but what AI thinks is the most engaging to me. I have some youtubers which publish irregulary, and youtube never notifies about rare content. It feeds me by short videos, girls in bikini in thumbnails and I feel like it changed a lot last year.

[@chfr]

6:43, makes me murderous whenever I realize an "article" is written by "company that JUST SO HAPPENED to sell you their solution"

[@jaredteaches894]

This is actually incredibly common, even before AI became an issue

[@funicon3689]

it's called content marketing

[@chfr]

@@jaredteaches894 I know about SEO slop, it was a pain back then and now it's even more of a pain

[@TheHubra]

Yeah! This is not strictly an AI problem but its made it harder to detect as you're browsing.

[@mihailazar2487]

I know **exactly** what you mean. Normally im as peaceful and tolerant as an Air bender but seeing crap like that... And not just seeing it but having to go OUT OF MY WAY to avoid seeing it, WITH LIMITED SUCCESS... it makes me CRAVE VIOLENCE

[@--ACCEPT--]

That visao article reading really struck something. I'm in my early 30s but I often work with people in their early 20s. Something I've noticed is that these younger people often write exactly the same way as this article is written. They often use the simplified grammar and passive voice of an instruction manual in prose, skip over details where they would be necessary but add (invent) them where they aren't, use technical terms mostly interchangeably (Is it a file? Is ist a file format? Who knows?), don't check their writing for plausability etc. I've always suspected that I'm just coming into my "kids these days" era, but maybe it's a case of "You write what you read". Maybe for these young people, this SEO Slop writing is what they associate with "Useful instructions", because that's what they've been getting served whenever they try to find any kind of information

[@ZedaZ80]

Recently had a similar experience, I assumed they used an AI writing tool with typos enabled.

[@--ACCEPT--]

@ZedaZ80 Yeah that would have been my first guess too, but without going into detail too much, I would have known for certain weather or not they did. Also, to be honest, some.of the writing was just too bad in other ways to be AI ^^'

[@tailwindmechanics7454]

Or the AI training data contained an inordinate amount of student academic papers, thus creating a bot that writes in that style. It's worth noting the lower cost AI models used for this type of slop are not the same "premium" ChatGpt/Gemini models

[@RetroRedivivus]

Good point.

[@jasongarcia2140]

Don't ever sell yourself short by just thinking you're in your kids these days phase.. You will always be a wise intelligent being. Don't give in to that stuff. These are real things happening to our society. It makes me so happy to know that people like you notice these things. They are quite glaring really right?

[@felixinfinita3777]

Here's a 3 simple steps how to convert a diversion valve into a PDF files formats: Step 1. Locate your diversion valve (windows) Step 2. Open diversion valve free online conversion program on the web Conclusion. Diversion valve is the future of the storage and transportation of the water fluids. So let's see what you can create with the amazing new tool, embrace yourself onto an amazing valve journey!

[@XerShadowTail]

Hi! I bought one diversion valve (windows) from Home Depot and I am wondering how a diversion valve (windows) can help me improve the productivity of my businesses. I used the online diversion valve (windows) conversion tool and was amazed at how it improved the productivity of my business. Here is how you can improve your business with diversion valve (windows). How can embracing my valve journey create improved business productivity?

[@Gold753]

peter-griffin-straightjacket.gif

[@epicman9105]

(windows)

[@noisetide]

@@XerShadowTail 💀

[@transhumanisttv1771]

This is just wrong. The thagomizer has to first be attached to the lingam and the thurim. The diversion valve will not function without the turbo encabulator. You're spreading false information.

[@Eva-ez1ks]

I'm a software engineer and usually when friends or family ask me about my thoughts on AI I reply with some version of "neural networks have some cool legitimate uses in approximating prohibitively slow/resource-intensive/unknown algorithms, but are not intelligent and the hype about AI right now is mostly misplaced". I have lately been very frustrated about the proliferation of these AI slop websites whenever I try to search for literally any information. This video has gotten me to think about how big this problem really is, so from now on when people ask I will sound the alarm about the deeply antisocial and destructive nature of generative AI.

[@cookie_plant8992]

I think text-AI is intelligent, just not really that usefull, because really stupid (you have to have intelligence to be stupid.) Slop requires intelligence and amorality to create (also greed). But mostly I would criticize google, facebook and co for allowing sub-par content on their platforms and promoting them via their algorithmic feeds.

[@SahilP2648]

AI in different fields is working in a different way. I don't know if you have been living under a rock but projects like Windsurf, Github copliot and Cursor allow you to develop at a much much faster rate than without. This is going to lead to layoffs in the tech industry and I don't see how it can reverse.

[@Fabojuno]

@@cookie_plant8992 It sounds intelligent but it is far from it. It just look refined because o billions of input text from real intelligent people (and some casual things so it can have more use-cases). what we haev is only an advanced autofill suggestion, based purely on use data and not in meaning. Business are only incorporating it because of pure marketing, they know this generative thing isn't agreggate in nothing, they are doing it for the capital sake.

[@bjni]

also a software engineer and I LOVE the LLMs they are so extremely useful for learning, googling and sorting stuff its insane, I dont use them to develop but just help with all kinds of basic tasks, sort through documentation present data nicely, do all kinds of small tasks that help speed up my learning process and workflow. Im cool if others dont like it and choose not to use it, thats fine too.

[@fredesch3158]

Right? In the past I used to try in shine light in the "pros and cons", like highlight why it's so bad, and how it doesn't work how people think it does, but also show it has applications, sadly people do not understand nuance (look at the replies you received in this exact comment to see exactly what I'm talking about), and people who are far from both the problems, applications and development of AI are even more immune to nuance in the discourse, it's either trash or the future, the "it has applications and probably could create cool things if used correctly, but people are using it to generate trash and destroy human work, while also using it completely opposed to how it was supposed to be used and neglecting actual research on it" flows over their head, they'll read into that reply whatever their pre-conceived bias tell them to. This video basically settled for me, if people won't understand nuance, I'll not add it to my discourse, if anyone asks me this is trash, any amount of work we can put in to slow this AI stuff is fair. The worst part is: even proper applications of AI are being obstructed by these stupid applications that worsens everything.

[@HimikkoStar]

5:22 "nobody writes like that", I know it may sound silly, but there's a reason people write articles like that, the Google search engine promotes content that has the keywords you type in, basically repetition of words like "format" or "file" or "file format" or "file extension" make your article more likely to be on the top search websites that appear when you search for that file extension. It's a strategy to have more flow of viewers in your page, and if it has ads means more money. Believe it or not this is intended, and it's written by a human most likely that actually was hired to write articles like that... I get your point though, it kinda feels bad, but that's just how this works...

[@jennysquibb7440]

@@HimikkoStar I don’t think this just “kinda feels bad”. Any person writing that way is writing for the machines, not for people. That is incredibly unfortunate and cross purposes with people trying to find good information :(

[@gi66itz]

Just because this is what SEO wants doesn't mean you can't do it more artfully. There is a balance being struck here between readable language and keyword overloading and keyword overloading is winning. I've working in advertising for many years and the copywriters who could do this well really deserved their paychecks.

[@MrPoeTayToe]

im a young computer engineer in a tech-heavy research position, so i consider myself relatively tech-savvy, but the examples you gave at the beginning really made me realize ive been reading more AI generated BS than I thought. while googling many obscure file formats used for my job and just quickly parsing pages to find the info I need, ive seen "this up and coming file format that is becoming super popular! its the best choice!" so many times and never really thought about it. i just assumed "hmm well thats cool, good thing im using this format if its so popular", now i'm almost 100% certain it was just AI marketing. thank you for making this video freya! the description got a laugh out of me, despite the sadness of it's accuracy

[@humanharddrive1]

It seems people are starting to suspect something... Most engineering jobs is fucking fast-fashion in tech form. Coding is not physics, where nothing has changed since the big bang and people have to discover everything. It's not a law discipline where all of the processes of a country are based on those fucking laws that change very slowly and require you to parse them. When you're working in a programming environment, you've got libraries, languages, products, constantly coming out and dying, and most of the time even the tasks themsevles have the same nature. A programmer is required to know a few basic algorithms, understand how programs get from text to execution, read someone else's code and fix what needs to be fixed. Most tasks come down to: find information, understand information, apply understanding. I really doubt that you're as tech-savvy as you think you are.

[@kaminekoch.7465]

Those examples started to become a problem like 2 decades ago. This is nothing new and there is a good chance those pages are older than any GPT model. Sure it's easier than ever and you don't even have to pay anybody in Pakistan to make these anymore, but this specific example is a problem with SEO, not AI.

[@AnotherAustin-z7b]

Yeah it's literally always been like this. Don't you remember the enormous confusion around gigabits and gigabytes? Probably too young for that but you will still find people screwing it up and getting it backwards and that will continue to the end of time

[@AnarchistArtificer]

I hadn't actually seen the video description until I read your comment and went back up to read it. It gave me a good chuckle, and I'm also enjoying the somewhat meta notion that this laugh of mine feels attributed both to Freya (as the writer of the description), and you, for causing me to go read it. The magic of human connection, eh?

[@AizakkuZ]

@@humanharddrive1 He said “relatively”. None of that has to do with applying critical thought to the sources you are reading except debugging. Totally different things that could mean he isn’t tech savvy, or it could just mean he’s admitting to being human.

[@CryZe92]

Only halfway through the video, but I can definitely recall that searching for these niche file formats quite often ended up in these very "templatey" / "free online converter" websites even before AI. But I guess now it's harder to tell them apart and there's probably more of it.

[@EmmanuelMess]

Yes, I've learned to search "around" those

[@capdyn735]

I know even in the 2010s I would search \*odd file format name\* + Reddit/Wikipedia/GitHub to avoid all the useless converters.

[@UtterlyMuseless]

@@CryZe92 Low-effort, template-based file format sites have been around for a long time (though they used to be at least marginally useful and less verbose), but I think the fake blog post is a newer vector. You would also usually get at least some useful documentation on the first page of search results, right?

[@bitfield-185]

My thoughts exactly ^ most of these problems were actually pre-existing in some form or another, but AI has certainly amplified them quite a bit.

[@seowebua]

exactly, she's so surprised like if it's something new

[@Dxpress_]

18:58 This is the thing that irks me about every "tech company" website nowadays - they never explain what the heck their product/service even is. Every company "creates & delivers experiences", with a vision of "empowering you to reach your full potential", by "increasing your engagement & growth" and "optimizing performance", with some charts, analytics, and numbers about... _stuff,_ and fancy background animations of _things_ just... _vaguely happening,_ before eventually asking if you want to sign up. Sign up for what? Who are you? What is it you actually _do?_

[@xspager]

Google got so bad this woman considered religion.

[@hexxt_]

its a dude

[@kopsha]

What, in the name of god, does this even mean? 🙈🙉🙊

[@laden6675]

do you not know who this is or...

[@bare_gamer9128]

@@laden6675a graphics programmer?

[@aventurileluipetre]

are you OK?

[@myrife]

I work in marketing and your assumptions about people in business are accurate lol

[@XeZrunner]

If we think about it outside of a business perspective, a bunch of these companies trying to make money to the best of their ability above everything else really feels like when a kid is being childish and tries copying doing something that they have no real meaning over yet - they're just doing it for the sake of doing it, because that's what they saw. Much the same way, these companies will do anything to get that money and grow, seemingly infinitely, very often without ever actually ending up caring about the work they do or the passion behind it. Perhaps some companies started out caring, but then they'll go public and the whole narrative changes.

[@AlanTwoRings]

He who sups with the devil should have a long spoon.

[@clray123]

@@XeZrunner When it comes to AI slop, we're talking about one-"business"man run sweatshops in India, not real companies. These people are simpy desperate for money, their business is a bit like running a digital prostitution racket. They also don't really care about their "customers", they care about somehow exploiting the system to generate a small profit in a dog-eat-dog world which surrounds them (because of overpopulation and global politics).

[@guard13007]

Reminder that Luddites only wanted to be treated fairly, not hold back progress. Luddite should be a proud term for progress, not a pejorative for regression.

[@BosakMaw]

What's tragic is how youtube used this anti-AI video to push AI startup adds on how I can earn passive income with AI tools, use my voice to 'sing' any song to my AI generated girlfriend that agrees with everything I say and then use AI to write a book on any topic I want and call it a day.

[@RolandoGarza]

I got recommended AI tools to create artwork for children's books "with no prior knowledge". 😢

[@Curt-0001]

What's tragic is not having adblock installed.

[@NGC1433]

Why do you watch these ads? Get an adblock, they are free and super effective since forever. I haven't seen an actual youtube ad since 2012.

[@goodlookinouthomie1757]

I've perfected the art of completely ignoring KZbin Apps. I suddenly become very interested in my ceiling or what's out my window for that 5 - 15 seconds.

[@BosakMaw]

To those concerned I have an ad blocker on my computer but I was watching it on my phone through the KZbin app, because I was on the move

[@t0rg3]

5:22 “nobody writes like that” - Trust me, I have been asked to write for a target audience where this style of writing was apparently required.

[@squishy-tomato]

yeah, we've had articles like that generated by people who don't understand what they write about long before llms, for search engine optimization

[@NavJack27gaming]

yeah this is completely SEO writing and not AI writing to me.

[@nixellion]

Oh yeah, I was about to comment just this. I've seen articles like that written years before generative AI was a thing. It was always so weird to write, like "who the hell wrote that? Who the hell writes like that?". So I would not blame it all on AI, and I actually even suspect that about half of the articles marked as AI may have been written by humans who just don't know a thing about what they write, and just dont care. Or maybe it's some kind of language barrier thing. So I agree that a lot of that could be SEO writing and not even AI. You could do another check to see the dates these were written. However that can also be altered, because for SEO you have to mark your articles as new, otherwise they may not show up on the front page. So many of these articles may have been written years ago, and just getting their timestamp updated.

[@NikkiDimesYT]

Agreed, there are far too many mistakes and oddities in this to have been written by AI. AI may hallucinate and make shit up, but it does so confidently and with perfect grammar.

[@terrrrmus]

@@NikkiDimesYT You can ask the AI to include typos.

[@welldias9668]

In portuguese we have this slang called “lero-lero”, which means to say a bunch of words and yet it says nothing. That is exacly what text generator AI does, “lero-lero”.

[@henriq031]

Esse tal de chat gabriel pedro tavares só escreve lero-lero

[@gelopisaminasotaripitanasi5921]

Does it translate to „empty-empty“?

[@AluRooftop]

@@gelopisaminasotaripitanasi5921 It's onomatopoeic, it'd probably translate to blah-blah

[@darksidegryphon5393]

In Portugal it's "encher chouriço/linguiça".

[@floatingmaster]

It's hard to distinguish between a human generated SEO-optimized text and generative AI nonsence...

[@firefly618]

I agree, they should both be equally banned from platforms like search engines and social media.

[@tommelling9847]

It's a circle, the AI's are trained on SEO slop and then regurgitate the same slop. When they try to train it on actual blogs and forum text you get results like adding glue to pizza to stop the cheese from sliding off :D

[@ske2004]

i ignore both

[@andrewdunbar828]

AI slop has a lot fewer grammatical errors than traditional generated slop on the web. I think the slop that's the subject of this video is not AI slop.

[@quantum_ocean]

don't worry, soon it will be clear because human-generated comments will be all the dumb nonsense ones.

[@Elyaradine]

One of the worst parts, I think, is that even if we curated something that was very human, a nice, pristine place where this litter didn't happen, it would likely just be used as a goldmine of non-slop for feeding more AI trash.

[@blockalism]

Another of the worst parts is that the more work you do to be inclusive of all humans - like putting ALT text on your images and transcribing your videos - the more you'll get stolen from, because that data is vital to AI training.

[@AileTheAlien]

🤔I hope the search engines could somehow train a AIs to filter out slop...but this is probably a recursive nightmare problem to solve. 😑

[@qaztim11]

Glaze and nightshade everything you post online, the tools to fight against generative AI exist, but need more adoption in general.

[@SovermanandVioboy]

​@@AccountHolder-c5e And your job will BE replaced too 😉

[@AKS9]

It seems that one solution is invite-only online communities where the content is behind a login and can't be scraped. Being invite-only, with say a limited number of invites possible per a given time period, would help ensure that nobody who has the intent to scrape the content would ever be able to become a member. I know there are communities like this that exist, for example on Mastodon and elsewhere.

[@lnx0007]

I asked chatGPT "write a tech blog article about .GLB files" and what it spit out was indistinguishable from any of those blog articles :/

[@elrichardo1337]

dead Internet theory rings truer and truer every day 😭

[@quantumgaming9180]

It's not a dead internet theory, but a dead internet reality

[@swojnowski453]

that's what happens when you believe search engines that tell you that you should not link to other websites. People who own a website stopped linking to others. SEs have become centres for search rather than source websites. Then the SEs started spitting AI stuff mostly.

[@AximVidya]

@@quantumgaming9180 it's using theory in the scientific sense, i.e. the best available model to describe reality

[@dievas_]

For me internet died when social networks dominated it, before you had to join some forum, where enthusiasts shared very valuable information, it was golden age of internet, information was of a very high quality. Social networks killed all of that, most forums died, information quality was deteriorating fast, because of how social metwirk algorythms worked - social networks are designed to keep you there for long time, not find quality information fast. Starting with cooking communities and getting to hacking and bomb making communities on the extreme end. Generative AI is just kicking a dead "body" of dead internet.

[@SterileNeutrino]

Neal Stephenson hinted that this would happen in "Anathem" “Artificial Inanity systems of enormous sophistication and power were built for exactly the purpose Fraa Osa has mentioned. In no time at all, the praxis leaked to the commercial sector and spread to the Rampant Orphan Botnet Ecologies. Never mind. The point is that there was a sort of Dark Age on the Reticulum that lasted until my Ita forerunners were able to bring matters in hand.”

[@mr.boloso8296]

When I was in high school, teachers told us not to use Wikipedia as a bibliography because the content there might be unreliable. Now it seems the other way around. Well, I'm curious to see what will happen when the internet is flooded with this AI crap. Will we embrace tradition and read books? Will they also be AI generated or become obsolete? Who knows! Great video, Freya!

[@dudelookatree]

There are published books of AI slop

[@abdul4515]

@@dudelookatree Yeah, it's a common get rich quick scheme. I've seen KZbin encouraging people to make AI slop books just so they can profit from it. It's gross.

[@SterileNeutrino]

Books have been subverted by DEI ideology, keep the original of Lord of the Rings.

[@1495978707]

​@@SterileNeutrinoNo, that's wikipedia. Books have a barrier to entry that keeps a lot of the virtue signalers and ambulance chasers from bothering

[@abdul4515]

Only reliable sources will be research papers at that point. And let's hope that Universities make sure that no one uses AI to make those either.

[@celdaemon]

I seriously just want AI to disappear overnight, that's my #1 wish right now. Every time these big companies do stuff like this, the internet becomes less about what it was actually intended for, people. I want interesting things here, written by people, for people. Sure, automated tools will always be used, and SEO spam will always exist, but this AI slop is just awful.

[@mahiainti678]

21:00 "why did they add that part?" they are clearly gaming SEO with all those keyword-filled lengthy filler paragraphs...

[@fionaskittle]

You've put your finger on exactly the same dread and disgust I feel about generative AI. It's horrible, I hate it, it's destroying everything I hold dear.

[@asailijhijr]

Some of this slop is SEO heavy, which makes it hard to read. Some of this slop is templated, which makes it difficult to scan (with human eyes) for useful information. Some of this slop is slop, which makes it just awful all over and throughout.

[@ingoverhulst]

I work in education, and the point you made at around 39 minutes - about it all being deceptive- is what bothers me most. Students have always tried to BS their way through reports, and in the past I could quite easily flip through their work and know whether it was garbage or not. Now, however, I have to read everything in full - especially if they have put in some effort to rewrite parts of AI generated stuff- before I can make my BS call. My workload has increased by at least 100% in exam periods because of this, whereas students probably spend less time on education than ever before. As a result we're going back to good old-fashioned on location exams, instead of doing-what i deem to be way more value adding- case reports. Had to get that of my chest, now back to the video.

[@user-wo5dm8ci1g]

The most interesting take I heard a year or two ago (on the TrashFuture podcast) was this: "AI is the anti-printing press". Basically, the printing press facilitated remote human-to-human communication at an unprecedented scale. And AI, at least in its ideal form, removes human to human communication. How do you find real people, with real ideas and information to share, or communicate with those people, when AI is flooding all the communication channels with imitation communication?

[@walter0bz]

People who want to find real people can look for channels like this one and filter for them . A dedicated curated authentic human search could be built.

[@SioxerNikita]

What the hell? The Printing Press was not "human-to-human" communication. It was the first time where it heavily encouraged not being particularly personable with the reader, because you could MASS produce text. The Printing Press made personal letters just a bit more redundant, you didn't have to send letters, and receive them, to get the basic gist of what is going on, because newsletters became a thing, because it could be mass produced. It was quite literally the first step towards removing personable human-to-human communication. Of ALL!!!! the examples you could give, this one is the absolute worst one you could've chosen to go with, even if you heard it elsewhere.

[@diskgrinder]

@@SioxerNikitano need to be so snippy. Especially as you seem to have missed the point.

[@SioxerNikita]

@@diskgrinder No, I understand the point attempted to be made, but the medium to do so is terrible.

[@mushroomdude123]

@@SioxerNikitaExcept the invention of mewsletters didn’t come about until the internet. Books _are_ a form of communication and the printing press made it easier. You’re being obtuse for no reason.

[@ivanjermakov]

"file format" is a very known pattern to leech on for such throwaway websites. "gltf glb layout" gives concise results with links to Khronos and Wikipedia.

[@ficolas2]

Yeah, these sites have existed way before AI

[@mattmmilli8287]

I had to parse a glb and repack with some additional stuff once in a JS web worker x-x learned so much on that one. The very detailed info definitely all out there even 5 years ago

[@LDVSoft]

Yeah, SSO leaching and LLM generation together really create poisoned lands.

[@jmvr]

I usually tack on "specifications" to get good info about the file format, such as "GLB format specification". It usually takes me to a detailed guide on how to parse the file format, and has worked for me when trying to parse PNG, JSON, and other formats. It's usually a spec guide from the group that designed the format, and it's sometimes so detailed that they'll even give examples for what you _won't_ find, like with the JSON spec.

[@gameofpj3286]

"gltf specification" also works well btw :D

[@Riderbooker]

The thing that makes me even sadder is that no amount of impassioned speech is going to move the needle away from this bullshit. It legitimately makes me terrified of starting any sort of creative endeavours.

[@joshinils]

I feel like a crowdfunded solution is becoming necessary. something like sponsorblock, but for top level domains. which is then implemented as an addon that removes/highlights links to likely ai garbage. and you then also have the option to vote on a site as being ai or not (or other categories, for example facebook could be markes as including ai stuff) and trusted websites could also be set up, like wikipedia.

[@tlacmen]

Who do you expect to do the voting? AI obviously ... sad.

[@joshinils]

@@tlacmen I already have an addon which lets me block result urls in google. so i am already doing this, me, a human. (feels silly to say i am a human)

[@henrylindeman4499]

Isn't that basically how they train the models? RLHF - AI generates crap, humans decide whether it's AI or not, use that data to further train the AI until it's indistinguishable

[@joshinils]

@@henrylindeman4499 i don't know how much is actually supervised. And previously everything on the internet wasnt ai generated. Nowadays its becoming garbage in, garbage out. So unsupervised could be detrimental overall, yes.

[@AileTheAlien]

🙂This seems like it's at least worth a shot! Otherwise we all just get to deal with endless slop. 🥲

[@CrazyFikus]

That "discussion" at the end... words fail me. STEP files were originally developed in the '80s, and became an industry standard in the mid '90s. Calling them a gamechanger like they're a brand new thing that just came out... yeah... I think it's time to move back to user made forums. Dealing with a power tripping admin that had a nervous breakdown because of some inane and petty personal drama is preferable to this slop.

[@FoobarDesign]

It's full of nonsense. As a FreeCAD user I was cringing at the 'viewer with basic editing features' thing. FreeCAD is what I use daily for 3D modelling and while it's far from perfect (or even stable) it's actually very advanced for a free 'software program'.

[@sateeshmodukuru1]

What are you doing STEP file bro?

[@MajSmerkol]

If no one was bothered to write it, I don't want to be bothered to read it.

[@josephnewton]

Thank you for sharing the step file podcast episode, it's something I'd normally scroll past and not realise the ridiculousness of. Now my face is shrink-wrapped to my skull as a result of wincing so much.

[@acegikmo]

you're welcome/I'm sorry

[@SeanJMay]

The first time the feeling you are imparting hit me was ~25 years ago. I was coming out of high school and dreaming of making a career as an artist in the performing arts. Looking through general lists of casting calls, I ran across more than one casting call for an "all-girl" or "all-boy" band, which, not my thing, but fine... but the call was for dancers/models, for a band that ALREADY had their album written and recorded. They were hiring band members for music that already existed. The studios were just looking for the faces and names that would appear on the album cover, and dance to the tracks on tour, and be in the posters on the walls of all of the tween girls, for the next 12-18 months. I've spent roughly that amount of time trying to cope with the disillusioned cynicism produced by the realization that when the investors "understand" the market, and how to game the market, that they will do whatever it takes to manufacture whatever consent and/or zeitgeist to meet the ends of attaining influence and maximizing profit in that market, be it music, or ... guided missiles... Tail wagging the dog. Man behind the curtain... Cassandra's curse, as applied to "the invisible hand of the market" is what it is. The corporate commodification of machine learning; selling it as a way of replacing human artists... and human communication... and all of the rest, for the sake of captive audiences, and overblown "viewer" metrics to overcharge advertisers for... I'm not going to begrudge someone for using these tools to generate their own content, if their desire for passive consumption is that basic. I will absolutely take issue with the people who then insist their bland, statistically-middling content be foisted upon everyone else, at market prices, as the "pinnacle" of their creative capabilities. Especially from corporations who already did everything they could to put profit over human wellbeing... though the AI-bubble is tracking along just fine with the dot-com bubble, as well. Don't give up. Keep doing you. Punk was a counterculture response to the commodification of Brit-pop and Rock'n'Roll, until it was, itself commodified. Grunge was a counterculture response to the hair-metal and glam-rock commodification of the '80s. There will be a counter-culture response to this moment, as well... and it will invariably be coopted, like all of the others. But remember that the roots of the cultural shifts are always planted by the thinkers and the dreamers, and once the movement has been coopted and homogenized, it stagnates and withers, no matter how much the shareholders will it to remain stable, and ride the trend into the ground. It was true in music, multiple times. It was true in film in the 2000s and 2010s. We're seeing it in big game studios, today. AI as a means of cost-savings for all of these industries isn't going to prevent that stagnation, nor prevent the thinkers, nor dreamers, of inventing the next countercultural shift. You've got this.

[@kees_]

Wow fantastic anecdote

[@Eveseptir]

Best comment I've ever read. This is the world, you nailed it. Marketing with powers akin to a cosmic horror, blending human endeavor into a slurry and squeezing the juice out. Stay cool, life grows out of the compost.

[@RealFableFox]

dancers / models? are you talking about background / foreground dancers or what? or the band itself? in case of Idols - it is an industry in itself and everyone knows how it works. the same for Bollywood industry.

[@JamEngulfer]

Are you sure it wasn’t a call for backing dancers?

[@SeanJMay]

@JamEngulfer they weren't calls for dance troupes. Dancing isn't my forte, and I wasn't looking for calls for dance auditions. They were essentially "idol" positions in the slew of copycat boyband/girlband post N'Sync / Spice Girls. And that's perhaps too much credit, because in JPop/KPop, depending on the era, you might at least expect an idol to dance and sing at the same time... at least, capturing the performance during dress rehearsal, if not live off the floor. And this isn't the first time in history that it happened. Boney M existed because a German musician wanted to write disco music, but he wasn't taken seriously, because... Germany wasn't a big scene for disco writing... So he hired a bunch of Carribean idols, essentially, and a couple of women who could dance and sing. That's why the dude who sings Rasputin has a German accent, despite nobody in Boney M looking particularly German, based on press photos. The rest of the group was essentially the precursor to Milli Vanilli.

[@marcosgil5447]

Hey, I first want to say that I really appreciate the effort you put into making this video, especially as it regards a topic that seems to be very sensitive to you. It certainly got me into thinking about a bunch of things, which I'd like to share in this comment. There are three points which I’d like to discuss: arts, politics and science. Starting with arts, I just want to put some context to where these thoughts are coming from: I'm not an artist by profession, although, for a time, I was into writing music, and I consider myself a person that likes art. That is to say that the consequences of AI for art do not affect me the same way as it affects artists, so there might be some nuances which I'm not aware of. Having that in mind, I understand and agree with the concerns by artists that the existence of generative AI relies on their work for training purposes and is extremely unfair that these models utilize them without neither permission nor monetary compensation, especially as it produces content that competes and, many times, replaces their work. During a certain part of your video, I found myself in the terrible position of trying to define art and the relation of AI to that. For me, art is about novel ways of communication, of saying things, of expressing thoughts and feelings. As you said, the advent of photography, instead of killing painting, actually made a revolution, as people began to question if realism was really the only valid way of expressing things. In that sense, I agreed with you that AI generated art isn’t really art, because it is not communication, it is just a pretty picture generated by a computer. As you said, there is no soul. There is no one on the other end to establish communication. But, then, a memory came to me: it was 2020 or 2021, so Covid, heavy lockdowns and so on. A friend just recommended to me a story by Jorge Luis Borges called Aleph. It is super abstract and crazy and at some point there is this thing where the character sees the entire universe, all of human knowledge, all that were and could be, lying just under a staircase. That image really stuck into my head for days, and if I were a proper artist I feel like I could have done something awesome with that, but I’m not. Then I saw these AI image generators that were, back then, just popping up, and decided to prompt it to see what it produced: and it was perfect. As these models were just starting, they were, in a sense, horrible: low resolution images, not very realistic, prone to hallucinate things. But as I prompted it with such an abstract theme, these flaws were actually not flaws, it was just the style I was looking for, it really captured the zeitgeist. I then showed it to my friend, he found it awesome too, we got talking for a few hours about the story and the generated image, and it was a very beautiful moment for me. In this case, I believe that what the AI produced was genuine art: it provided novel ways for me to communicate deeply with a friend of mine. Of course this is a very particular experience, it probably isn’t representative of the role that AI generated art is having in most cases, but it is, in my opinion, at least a counterexample that AI is always a parasitic cancer, as you say. And I guess this illustrates my general opinion about AI in general: it is just another new tool, an extremely powerful one, nonetheless, that can be used for both good and bad, and it is not clear for me yet which use will prevail. To justify my view of AI being just another tool, I move on to politics. You correctly note the harmful use of AI to influence elections. But it is important to notice that people were already using social media to this end before the boom of generative AI as we know it. The Cambridge Analytica case is one of them, and I also point out the 2018 election of Bolsonaro in Brazil (my country), in which there was a heavy use of bots to spread misinformation. Of course AI makes this particular problem worse, but my point is that it did not cause it, it was already there. Just to finish, I’d like to comment a little bit on the relation of AI with science (my field). There has been this certain trend of wanting to put AI into every possible work, as it is very hyped and will certainly attract a lot of citations, which is, unfortunately, how scientists measure their value. I personally started a work using AI, but later found a better method and had a hard time convincing my collaborators that it was indeed better. I find this kind of behavior bad, because I feel like it kills innovation, as people start to approach problems with a very narrow field of view. Nonetheless it is undeniable that AI has solved extremely important problems that we do not know how to solve otherwise: an example is the Nobel Prize of Chemistry of 2024 awarded to the team of Alphafold, which is an AI model which is able to predict the structure of proteins. This has implications for drug development and biochemistry in general. To conclude, I just wanted to thank you again for your excellent video and to spark up this extremely important conversation.

[@cognisent_]

Two things: 1. You reading these articles, especially the first one, is probably one of the best vocalizations of how my brain works that I've seen. I had to call my partner in to watch it so he could understand me better. 2. All of these, every single one, is created specifically for Search Engine Optimization. Pages with more keywords will feature higher in more searches than pages written with informative, focused and concise content. It's the shotgun approach that has proven most successful in gaming Google and others, and which has directly led to a progressive "poisoning" of the indexes which manifests as "worse results". This was happening before (LLM) AI, but it's just happening faster now.

[@mousey796]

Haha, re (1) I just did the same thing with a friend and then saw your comment afterward. My people 😊.

[@realGBx64]

Imagine the text an LLM trained on these SEO shotgun pellets would write 😂

[@cognisent_]

@realGBx64 Model Collapse

[@SioxerNikita]

The funny part, the exact problem can also be the solution to the problem.

[@SioxerNikita]

@@cognisent_ Erm... The LLMs have already been trained on these... very likely... I don't think you really understand what is going on here, or what model collapse even is, or who is even predicting it, and we don't really have evidence of it happening.

[@hexarith]

Regarding that last part (the discussion): A coworker of mine fed my latest (end of 2024) paper to the tool making these "podcasts". And after superficial listening to it, I found so many misrepresentations in it. I was thinking (but not seriously considering) to do a video/talk/article deconstructing that generated output and highlighting what parts of the paper it's currently "discussing" and how it's unrepresentative, or puts emphasis on the wrong things. Shall I go forward with it?

[@sheargrub]

More quality analysis is always nice! Might be able to find an audience if you post it in the right part of BlueSky or the like

[@diedevanmarle]

That’s really interesting! But as a person that has worked on AI in academics (from a design perspective) pointing out how AI or LLM’s misrepresents information is quite difficult. This misrepresentation is not a general thing and often seems to rely on the specific LLM model. After all these models do not reason and merely utter words based on what words are likely to be used in conjunction with the words you put in. We see a lot of students learning how to do “prompting” so that they can get perfect results back from the model, only to discover that after a tiny OpenAI update, their prompts do not work anymore! It would be interesting to point out how the use of ai promotes these mistakes and how AI models (because they lack reasoning) generally respond unrepresentative results

[@jan_harald]

yes

[@neetfreek9921]

I don’t know it seems more like people misusing the tool rather than an issue with the tool itself. It’s not meant to do 100% of your workflow. Just a good chunk of it. You still need to edit it.

[@sheargrub]

@@neetfreek9921 In fairness, widespread misuse seems like a pretty good sign that people need to be made more aware of the possible pitfalls lol

[@Globox610]

I'm not nearly smart enough to comprehend or even comment on the dephts of this issue, but at the same time I felt like I had to say something. So as someone who's always been interested in the beauty of art and storytelling, this video resonated with me. I guess all I can add now that I've also started to dabble in arts and game making, I'll happily follow in your footsteps in learning for the sake of being able to tell my own story more than producing an end product.

[@jakubnezvaljn]

getting ads about AI generated presentations and videos on this specific video is wild and only underlines Freya's point

[@RadeonVega64]

yep

[@walrien9359]

Firefox + uBlock Origin, I haven't gotten a single ad on youtube while using that

[@SaHaRaSquad]

@@walrien9359 Yes you have, you forgot sponsorblock

[@DoChDev]

It is ironic and also quite telling, that the only use for Generative AI you found is helping you search the web. The same thing they made nonfunctional with their slop. Truly a companys dream, creating a problem and then selling the solution.

[@capdyn735]

I use it somewhat for my work (programming) and it helps for four things in particular: 1) Helping search the web 2) Acting as a glorified search and replace tool. Sometimes I need to change the way a few functions or classes work which leads to a dramatic change in the places they're called. If it's going to take me more than 5 minutes or so to change them all then I will copy all into a code block in a locally run LLM, then correctly change one and ask it to do the rest. Works like a charm and saves a couple of minutes. 3) Due to mental block or whatever sometimes I just cannot think of a good name for a variable/function/class, so in order to not spend forever coming up with a name or creating a Java-style one I explain what it's representing and ask for some suggestions. This usually saves some time. 4) I'm not confident in my ability to write concisely (see: this comment) so it's helpful to chuck documentation I write into it just to check for conciseness and clarity. These are all just using the models to transform language rather than in a generative way since I don't trust their output when asking them to write stuff based solely on description. For me it's just another tool, it works well for what I use it for but I wouldn't use it outside of that.

[@callyral]

"Wait, if everyone had a gun, there'd be no more mass shootin- I really gotta get out of Florida"

[@R.B.]

Google Search became unusable about a decade ago so I switched to Bing and Duck Duck Go. That's been in decline because of SEO spamming. When Bing Chat was first released two years ago, it was the best for search. Citations and good results. It's declined a little as they've added filters and blocked the ability to find some content, but generally I get my best search results that way. Gemini generally fails spectacularly, GIGO probably because of SEO. Perplexity for certain searches is also a very good resource. I don't know that everything shown is necessarily AI generated, but it is definitely SEO pollution.

[@8qk67acq5]

Eventually, it'll get much better.

[@R.B.]

Google Search became unusable about a decade ago so I switched to Bing and Duck Duck Go. That's been in decline because of SEO spamming. When Bing Chat was first released two years ago, it was the best for search. Citations and good results. It's declined a little as they've added filters and blocked the ability to find some content, but generally I get my best search results that way. Gemini generally fails spectacularly, GIGO probably because of SEO. Perplexity for certain searches is also a very good resource. I don't know that everything shown is necessarily AI generated, but it is definitely SEO pollution. KZbin comments are awful because most of my comments are hidden.

[@Dysiode]

Kagi is the search alternative I'm most familiar with being a paid (mostly) search engine that proclaims how that lets them be more fair. Using the same search the literal glTF spec doc on Khronos is the 20th search result, of the other 19 one is wikipedia, one is a reddit post, and one is a docs page for a javascript loader library, everything else is one of the websites you covered. And like, I get that "glb file format" is a very open ended search, but jeeze. Even adding "spec" to the search is filled with AI generated crap. It's like we're going to need a "was the page created after 2022" filter

[@lucyg00se]

Thank you. I'm sorry for the crisis of faith you went through / are going through. I'm glad I'm not the only person going through it. I work in academia in maths education, and I get the feeling more and more each day that I can't trust that any given article published after 2021 is entirely written by a person. Any time I attend a seminar that has AI in the title, it's a toss-up whether it's a "ok gang we reeaaally need to rethink assignment briefs for students in this new paradigm of hard-to-kill superplagiarism" or a "I used AI to generate assignment briefs and here's how you can too" (both of which would be titled "The impact of AI on designing assignment briefs"). I think what we can do to have hope is to share these experiences, like you've done, so that we know other people also deeply care. So thank you for sharing this.

[@Graham_Wideman]

It brings into focus some new objectives for education at a meta level -- examining what is even the role of personally-held knowledge, or how to research and reason in an environment riddled with AI-generated content, how to vet sources, look for internal consistency or consistency across sources.

[@thamass8772]

You're so incredibly real for this. Idk if it's because our personal interests around the internet overlap here but I find it very rare to find someone else from the technical side who shares the same thoughts about Gen AI. In game development I find people less and less oriented to this perspective and I feel like I'm in an upside down world where no one else can feel the water boiling. It's not a fear of new technology or worrying about job security but a moral and existential issue that erodes the potential of people. Thanks for taking the time to talk about this. Greatly appreciated.

[@sergeygladkikh8520]

Thank you, it’s important that people like you speak about this, and hopefully this video gets the attention it deserves. I‘m an AI optimist myself but I also share the opinion that we may have a very rocky transition phase, and the potential challenges like those you mention here must not be ignored.

[@gregcsokas]

Every of these texts are wordy and weird for a reason, this is the mighty science of SEO, back in the days, when I worked in a company (20 years ago) where we did SEO we followed the same rules just manually. Be wordy, repeat the SEO keywords like shit-ton times. Any fillers in the text which doesn't make any sense (glb to png conversion for eg) is just raising the density of the glb keyword. (again, repeat the keyword shit-ton of times) but the robots don't understand anything about the context. They just tokenize and calculating a value for the webpage. So yes, it's shit for purpose. And that's why you found these garbage things on the first page, our mighty AI tech is wonderful for generating this SEO shit.

[@cookie_plant8992]

You are essentially saying that AI is really good at what it is supposed to do in this context, but that the systems we use for creating content (especially the platforms for media consumption) are not built to reproduce quality?

[@gregcsokas]

​@@cookie_plant8992 I'm saying, AI is great for generating input for search engine robots, to achieve higher scores. On the other hand, search engines are bad for delivering quality content. For e.g. opening a bad result gives some amount of validation to the search engine, but giving feedback is like 5 minutes of active "work" from the user side. And no one sending feedbacks to the Google about the results quality I guess.

[@roastyou666]

@@cookie_plant8992 They are bulit to increase chance of business exposure. Social media is an illusion for consumers to think they will get any quality content in. They are built with profit in mind. The only way to break from it is to self-host a fediverse application suite.

[@seowebua]

"just raising the density" doesn't work for like 10 years already

[@ColonDee.]

I'm looking forward to this. I must admit i wasn't expecting to see a video about this on this channel, out of all channels. It was a surprise, but a good one, as I too think generative AI is quite problematic (mostly for how corporations are conducting it) and I feel there isn't enough discussion about it, given its power. Edit: Omg, I barely made it through the "let's figure out the specifics of the GLB file format". It was so painful to watch because it brought back memories of very similar experiences I had not long ago. And this is indeed something new, that's not how things used to be. I really miss the times when you'd find a very useful forum thread or blog post from real people dedicated to a subject. Also, unfortunately this isn't limited to just written content and Google searches anymore. But honestly, I'd go so far as to say that I feel this not exclusively an AI thing. The internet has been year after year becoming less organic and more corporate. It's now a place for products and (potential) consumers, in contrast to being a space that people build themselves. Generative IA is just an additional (super effective) actor on this whole process. I've felt a grief and a disinterest on the internet from years before the boom of genAI and today it doesn't effect me as much. But I appreciate what you're doing because if we're not to have a soulless internet, people have to discuss this problem. Even if genAI is just too good and is here to stay, at least its cons have be in the line of fire. (I wish I could share my thoughts on every single point you made in the video, cuz maaany of them felt so close to some of my concerns... but well, can't do that in a KZbin comment)

[@acegikmo]

I'm glad! it's scary releasing this, bc there's multiple things working against me ;-; 1. this topic is outside of what I usually talk about 2. it's a lower effort yet edited video which I usually don't do 3. it's about a pretty controversial topic, one where a lot of even my own viewers are likely to disagree with me

[@aurele2989]

​@@acegikmo curious for your take on the topic! "lower effort" how?

[@expired___milk]

@aurele2989 I'm guessing lower effort compared to her past few videos where the quality was through the roof.

[@jortor2932]

​@@acegikmono ur viewers are like u they love u

[@ColonDee.]

@@acegikmo Just watched it. There might be people that disagree, that's inevitable, but the discussion is valid and you present it in an interesting way by being very practical. I think I'm with @jortor2932 on this!

[@qeshmer]

I don't usually watch long KZbin videos but I had a hard time closing this tab because of how well it captured and echoed my thoughts. Your segment on depression and erasing humanity resonated with me very deeply. I've felt so much frustration with how so many people around me like coworkers, friends and the odd relative, find this technology "cool" without dedicating a single thought to the consequences it's bringing today. Any attempt at criticism from me is just dismissed. "Everyone is excited about this. Why aren't you using this technology?" This shit has poisoned so many of the areas of the web I get fulfilment from in these last few years, especially finding new art. I get so vexed when proponents and developers of these techonologies talk about potential doomsday scenarios of the future, as if the negative effects of today are just footnotes. I'm honestly not sure how I'm supposed to look forward to the future when AI is being shoehorned into every single product, especially when our lives are so tech dependant. The one thing I'm holding out hope for is that we will have strictly anti-AI creators/artists to rely on so there's still human-to-human sharing of experiences etc.

[@RetroRedivivus]

"I get so vexed when proponents and developers of these techonologies talk about potential doomsday scenarios of the future, as if the negative effects of today are just footnotes." Great point! I have also viewed the Holywood inspired AI doom porn as completely unnecessary, ridiculous and suspiciously convenient for anyone seeking more power who could use this latest fear to pass even more government regulations and ramp up "security" due to the potential for AI assisted cyber terrorism or cyber crime.. leading to more police state Patriot Act nonsense. But you brought up an angle I hadnt thought of. The extreme and actually impossible scifi threat from a robot uprising like in the movies could actually be having the intended effect of distracting the public from the very real and current problems and threats posed by the simple tools people are already using right now.

[@infinitelyexplosive4131]

When I was a kid I spent hours on the internet learning random things, clicking random links, and exploring. I have my job now because of the things I learned then. What do I say to the curious kids I meet today? Where can they go to learn?

[@carultch]

Paper encyclopedias. I browsed World Book for fun in the 90's.

[@yautl1]

Back to the local library, I guess, assuming it hasn't been defunded and closed down yet.

[@prokras8609]

To Discord, I guess. Kids are smart, they will adapt.

[@goodlookinouthomie1757]

More to the point how do you get them to watch anything other than some hysterically gibbering Minecraft streamer.

[@poepin3661]

ChatGPT or Claude, duh

[@voxelstack]

You do a great job talking about the main thing that makes me hopeless: AI is amazing at creating low effort garbage, and low effort garbage is what the people dumping money into it actually want. The worst thing about AI is actually the very thing people with the money and power to stop it have been striving for, and I don't see anything that can work against that.

[@Rubicola174]

It genuinely feels like there's a fundamental shift in how the internet works right now, and it's not all bad. AI-sloppification forces us to to be more carefull about vetting sources and will hopefully teach people to demand more reliable information. To immediately burn any source when they realize it's AI-generated. We are already getting plugins that block garbage AI-integration like Googles AI Overview and I fully believe it's just a matter of time until there will be plugins that ban websites presenting AI-generated information, similar to how spam filters or ad blockers work.

[@angeldude101]

AI really shows how big the flaws with capitalism are, and I don't see this changing without a massive overhaul to the entire economic system. Capitalism as it stands rewards cheap and rapid production that gets consumed quickly, and AI is simply the best tool for the job at that. It's just that the job in question is a massive perversion of the concept of human creation.

[@SkorjOlafsen]

Companies dumping money into making products that no one wants is a self-solving problem. There would be a real danger if people actually liked AI slop, but beyond an initial reaction of "it's cool a computer can do that", actual humans looking for actual products don't seem thrilled. So I predict AI will consume "corporate art": ads and product box art and the like, things we all try to ignore anyway. Perhaps it will challenge human-made slop in other areas (like low-effort "react" content) as well. Not a big loss. I think the real harm will come where we're forced to use already-low-quality services (like tech support call centers) which will inevitably keep getting worse thanks to AI. Areas where you haven no choice as a consumer are doomed to get even worse and more useless.

[@__cornflake__4252]

This year, I decided to go to community college instead of my high school. The school made me take an online english class for the last credit I needed. I'm convinced it was completely AI generated. The articles and videos I had to read and watch were all done in a robotic voice with ai generated faces talking over a slideshow of a mix between stock images and AI generated images. They repeated the same points over and over again. Sometimes the tests had questions about stuff we didn't learn. Some of the "medieval texts" were written in modern english. Near the beginning of the course, I started getting confused about why the "course" was teaching the same thing over and over again. I tried prompting ChatGPT for a short educational video script for a high school senior english class about the topic we were learning and it spit out nearly the exact same script as one of the videos I had to watch. It even used the exact same example from a book, along with the exact same, word-for-word disclaimer that the book wasn't british even though the class was about british literature. It was a horrible experience.

[@hexasquid]

1:17:09 hold on, I know exactly what went down! They had the cursor behind the "Notepad++" and tried to copy paste the entire text into the text-to-speech thing, but they accidentally hit "a" instead of "ctrl + a" once! idk, that's my best guess

[@almicc]

Something very annoying to me is searching technical problems and getting AI summaries/ responses and within two seconds finding obvious mistakes or incorrect information in the AI response. I would rather see crappy gif ads than a single sentence of an AI "guess" about very specific technical issues.

[@EugeneDubrovka]

Thank you! I've tried a AI helper for coding, our company provides us with one. First tries impress you, that's true. But then you pay with higher cognitive load. As if you are constantly reviewing the code of a trainee-student-junior developer. Basically it slows me down and makes review and not write code, effectively committing under my name and taking the responsibility. Kinda sucks! Turns out that AI-helper we use pays out only for some docs writing since it types faster than you and as a search for docs and examples with some context. It helps with learning new area but then sometimes just annoys: "write me tests for this class" -> produces stub with "// put your tests here".. Only "meatbag" is missing in that sentence.

[@keeblebrox]

This feels similar to the problem on the web that Google initially solved: finding relevant information. There's money to be made in the solution so there's some incentive to find it. In the meantime, we'll have to curate lists of useful resources on the web and share them by word of mouth.

[@diribigal]

Could we also bring back webrings?

[@aboringperson9069]

What you're describing kinda feels like the internet version of the Lorax movie-but instead of a company selling fresh air in a wasteland, it's a company selling bits of the internet that aren't plagued by LLM-generated slop. Hoping that another group comes along to sell us the old internet instead of just keeping the old internet feels wrong.

[@RetroRedivivus]

@keeblebrox - Right but did Google REALLY solve it the original problem of finding the most relevant information? Or did they find a way to simulate doing that well enough to fool most people? According to a few who suspected otherwise and tested their theory Google actually scammed us all. Where it tells you it found millions of search results for example if you actually click on page after page you eventually will reach the end but it wont be after you page through millions of results, it will suddenly stop showing you results at some point loooooong before it gets to that huge number it told you it found and it wont even explain what happened or why it wont show you those thousands, hundreds of thousands or even millions of results it claimed to have found. And in the process of "curating" results for you do you think you can trust them to be truly objective and to show you all the results you could possibly find relevant? Or is it more likely that at some point their own opinions, biases and beliefs are going to come into play and skew their curation methods, not to mention the fact that it is a company who's bottom line is profit and not actually to go above and beyond the customer's expectations. If there is a way to meet the average customer's expectations without any serious repercussions as far as their bottom line is concerned they are going to choose that way. And there is even the question as to which customer they are really catering to and prioritizing. Obviously its going to be the paying customers. Are you paying for your search results? Who's paying? Advertisers. So the real customers that Google or Alphabet seeks to keep happy is the advertisers who could care less if you get the best possible search results.

[@mateojuvera4691]

I know I am just a random stranger in the internet, but as a fellow programmer/shader wizard apprentice/creative on the side, I must say: Freya, you are one of my heroes. Like, for real. Your work and voice in this space are a constant inspiration and it makes me so happy to see you raise your voice and take this stance against a corpo-pilled world that seemingly does not care about the value of people's humanity. Your dissection of the problem is excellent and will be very useful whenever I need to explain to people why "AI" is cancer. (Although that is to be expected from how skillful you are at explaining complex topics). I can only thank you for all the effort, physical, mental and emotional, that you put into this and all of your endeavors. I don't know what the future holds, but I can assure you one thing: you do not stand alone. Many of us, whether touched by your work before or not, agree wholeheartedly and will much rather push back against the enslopification of media in favor for genuine human creation. At the end of the day, humans are a social species, so I can't think of many things more unnatural than trying to get rid of all the ways we have to genuinely form connections with other human beings. I hope that feeling of connection and human belonging keeps us strong and resolute in the face of this beast.

[@stumbling]

LLM's are that one guy who volunteers to do everything despite having zero experience.

[@RetroRedivivus]

And GAI is that one guy you would get if you took every guy like that in the world and plugged them into the Matrix and had them all coexisting in the mind of one Matrix avatar.

[@iandyke4412]

Topical: en.wikipedia.org/wiki/Kessler_syndrome When enshitification goes exponential.

[@joshinils]

the only solution is knowing what is man made, and what is not. every comment, every piece of content needs a certificate, and ones without get flagged. but changing the internet so fundamentally won't happen i fear

[@iandyke4412]

@joshinils i mean sounds like it require basically the same things that get talked about with deanonymizing everything. Maybe good for like trusted publications and such but theres the natural drawbacks brought in with that sort of system too

[@joshinils]

@@iandyke4412 yup, where does one begin proving they're human. Your government and or official passport could be one such place, but i don't want that attached to everything online. Not to mention how much could go wrong.

[@hypertectonics7009]

@@joshinils time to go offline, and stay offline...

[@DemonixTB]

hypertonics7009 psyop bot comment these "just go live in the woods" comments are everywhere once you start looking for them, and theyre always vague, always off mark, off topic and only exist to spread dread that keeps people complacent. if you are real stop and reconsider what you're spreading. But I will bet that you are not.

[@jugger017]

Web searches have been taking a nosedive in quality for the past 10 years, but it's really gotten crazy the past 5 with all this AI garbage. I want to go back to 2009 so bad

[@balthazar259]

The fact I kept getting ads for bs AI tools and products throughout this video makes it even more depressing

[@suanzy]

Indeed😂

[@arngorf]

I am an ML engineer (I don't like the term AI), and I recently, finally got to work on something other than "AI", and damn I hate AI slop, would never create such crap myself. We have been entering a period where we see a lot of chaotic approaches to try to earn money on this new technology. It is chaotic, it is brutal, it is impressive, it is scary, it is disheartening, it is ... a lot of emotions. What I am most scared about is how brutal this transition period will be, I'm not necessarily too scared about the end result, though that does also cause some level of fright. Zooming out a little. This "transition" will take a looooong time. AI is highly over-hyped, and that is a bubble that will either burst (as it has done 2-3 times previously), or we will see a slow down and investment drying up. Just like AAA game studios are facing a crisis today, so will the funding for AI slop dry up, leaving some of the best uses for us to work with. In the long term, I think we are swinging towards a Wall-E society, but I do think the realization, that being creative is fun, that people want to communicate with people, will happen eventually. AI is not as impressive as most make it sound, and the increments in performance that they (openai, google, etc.) need, are exponential, and their efficiency optimizations are lacking way behind. They are trading money/energy for better results, by doing the same thing many times over as seen in the new o3 model, and it is not sustainable, it will slow down significantly. I do a lot of programming, and the current best LLMs are orders of magnitudes away from solving the problems I solve, so that does not scare my. All the showcases are always, start a project for a calculator, or write snake, or generate a beautiful image of a river as the sun rises in the horizon. What I needs is "show me how to share memory allocation between zig and python, having zig control running the main function. I want to share allocated memory between zig arrays and python numpy arrays" and an artist using a generative model will have to ask something like "take this specific image of my character model, and place it standing behind a stool looking at this other character models hand, they look surprised, with a hint of sadness. In the background is X, lighting is like Y, flooring is like Z, etc, etc, etc"... They cannot do it, not even close, and they will not be able to for a looooong looong while. So I think we can chill for quite a while. The AI slop we see will undergo a continued, but probably short, period of hype, with "impressive" results, but then it will fade, because everyone will know it is only the 80% of what was needed, the last 20% being impossible to obtain without a real artist in the back. Cheer up Freya

[@arngorf]

A couple of rebuttals: AI destroying search engines: this is a capitalism problem, and it started before AI with 80% of googles search results being sponsored links in some way. AI has accelerated their destruction because ChatGPT in many cases is a better drop in replacement for subpar sponsored search results. The search engines are killing themself, and someone is bound to try to solve this problem, possibly with a new search system indexing the internet differently, but it is not invented yet. AI may create a virus that will kill humanity: It is technically possible that this will happen. I hope for a small event of this kind, which will show as a warning to the world how dangerous this is, but again, you cannot hallucinate a virus, so this has to come from something like Alphafold, which is not generative AI slop, and which is being used more carefully. I don't think this is an AI problem, but a problem of technology improving to a point where it becomes dangerous. Irrespective of AI, we may be moving towards one of the big filters hypothesis in cosmology, and that really scares me. AI may be accelerating it, but it was bound to happen anyways.

[@NLPprompter]

if you don't like they called AI then what you like it be called?

[@arngorf]

@@NLPprompter The AI terms is a hype term, but I otherwise don't mind its definition. I work in Machine Learning (ML) and other related fields, and not in AI, but if I want to get a job, I better call it AI, even if what they ask for is ML or Data Engineering or Image Analysis/Porcessing or, ..., and that is my issue... I was just referring to myself :)

[@clray123]

@@NLPprompter AI would be ok term if the "I" stood for "Imitation" (or "Interpolation").

[@jennysquibb7440]

Custom fitted and made shoes still exist and are valued. Mass produced shoes are very common. They are so common people could be excused in thinking that custom made shoes are not still a thing. Mass produced & standardized shoes provide the majority of what was needed. They also made it much more difficult for those that need something not provided by that standardization. I think it is entirely possible that AI slop is here to stay, that it will be “good enough” for the cost paid to create it, people will grumble about their options, the market for fully trained experts will shrink, and the training pipeline for experts will have a large chasm between professionals and those trying to get started.

[@Gobhoblin126]

Good to see the "advertising wrapper around open source tool" part of the internet has adopted AI

[@ninomojo]

I have loved every word you've said in this video. Especially near the end. You've managed to put things in a clear and well argued way, some of which I think a lot of us creatives were feeling but didn't know how to put in coherent words yet. You are seeing the bigger picture and it's refreshing.

[@johnadams6249]

I wish there was a way to flag search results as being bad, and to block entire domains from appearing in search results. It’s so hard to find high quality written documentation online these days. Everything is a video or is garbage text. You know a site is good when it has html styling straight from the early 2000s

[@Atropos148]

You can do actually do this* * but you have to pay for monthly subscription for search engine called Kagi I use it and have now blocked all the generated websites from this video

[@remmycat6913]

It's a paid service, but Kagi search does this really well (letting you block or prioritise certain domains) as well as a bunch of other processes and filters that make most search results actually useful again. And they use the money to also buy search results from e.g. Google, so you don't lose the actually good results and better processing of context that Google provides in comparison to other alt searches.

[@remmycat6913]

This sounds like an ad, but Kagi has mostly fixed web search rot for me, and I have accepted that you just have to pay money for a search engine to actually be helpful these days.

[@toboterxp8155]

I actually found a very easy solution to AI slop: Do not search for things in English. This is almost exclusively an English problem, because it's the most lucrative language to do this in. I have searched more in my native language this year than I have in the last decade.

[@watsonwrote]

That's great advice, thank you. I've been using English more regularly for digital things but I should try doing research in French instead to see how it goes. On y va !

[@JamEngulfer]

That’s so smart!

[@thbeam2044]

This comment probably won't make sense to anyone, but even let alone searching for things, recently I've been feeling like I can't _think_ things through properly in English anymore (my native language), and have to retreat to thinking in my very unfluent heritage language (Chinese) to think things through properly. I had no idea why this might have been the case. But, the idea that the quality of English material online is degrading, and by extension decreasing the quality of my own thinking in this language (since I conume so much online media in English only) is beginning to make sense to me.

[@absolutewisp]

It's really funny to me that this actually works now, since for the longest time I avoided searching technical topics in my native language due to the English results just being so much more recent and concise. Now it's almost flipped, since the English-speaking world had the misfortune of becoming prime real estate for AI slop ad revenue. Like, the Polish articles are still more out of date, but they're more useful 90% of the time.

[@CosmicSpiral_]

this is a good idea but it looks like these websites are just using computer translation to publish the same thing in different languages - in the example from the video, searching in french still gives pixcap, visao and so on, just in french

[@Valneal]

As a former artist, I’m not really afraid that GenAI will replace art as a medium of expression or communication. I think or at least hope people will come to realize that, like you, what they really valued about art all along was not the aesthetic surface but the human behind the creation and what they poured into it. Unfortunately we live in a capitalist hellscape where almost every aspect of life is commodified, including art, and artist’s livelihoods are very much in danger because our corporate overlords have created an attention-based economy where art is merely a tool to sell products, and that’s exactly what AI will be trained to be best at, while being cheap as dirt. Maybe I’m being a bit reductive, but I feel like it all just comes down to capitalism once again and the age old problems of as “cheap as possible while barely good enough to sell” and “if you need the product and everyone’s products are garbage, you better buy our garbage anyways”. AI is just turbocharging this to a massive degree, like any other technology that enables the system it is acting under, just more. It’s not surprising at all that the only useful result you got was a non-profit, Wikipedia. The “optimistic” thing to take away from this is that while AI is here to stay, capitalism can be replaced in principle. And I firmly believe that AI would take different shapes and be used much more for the betterment of humanity under a different system.

[@wkernelteam]

I think that this problem with bloating of search results is there for longer than chatgpt, maybe from mid-2010s it started growing progressively worse, especially if you don't know what you're looking for (like file formats, hehe). Some sites were full of SEO content, some were just copied posts from stackoverflow or documentation. But it surely was not as bad as it is now. All of a sudden I have to acknowledge that KZbin videos are becoming a better alternative to all this mess, even though text posts are supposed to be easier to make for humans, more time-saving. Bloated with ads too, but here you can still feel people being passionate. (Comments are mostly dead when it comes to viral and political content though). Thanks Freya for sharing this, it's really nice to feel that you are not alone in this thoughts.

[@lawrence4301]

youtube has also gotten worse like dramatically so, in basically every regard

[@hartvenus]

Yes, I’m actually convinced the first page is entirely written by basic web scrapers and outsourced humans to funnel people into their platform. Look at how every header is a common search query, and everything has steps even if not really necessary. This is all SEO optimization, which I ironically haven’t seen ai do. Also ai doesn’t really write with gramatical errors. The problem isn’t ai itself, it’s how easy to game SEO is.

[@ulidtko]

Every time big money comes in, it corrupts the space. The tragedy of GenAI is how *exorbitantly cheaper* it makes manufacturing SEO trash, at scale.

[@seowebua]

@@hartvenus wow, whitehat SEO is not easy and never was. Do you think other 9 mil results for that query didn't want to be in top1? "Headers with queries" are only the surface that you see

[@poepin3661]

Yeah seriously, web results have been filled with slop for over half a decade at this point. Thankfully now I have LLMs to sift through that slop for me ;)

[@JohnnoWaldmann]

We predicted this Ai nonsense a couple of decades ago. Especially problematic is false Ai generated copyright claims against content that was generated by real humans.

[@ReadThinkCreate]

There hasnt been a day that I havent thought about AI since ChatGPT was released. And it is hard to put into words the multitude of emotions I feel about it. But what you said in the second half of the video is a pretty good summary of how I feel---especially about what something human-made embodies. That was beautiful. Great video.

[@yv-odes]

i literally never comment on videos, but wanted to say thank you so much for saying this. i share a lot of your passions - and fears. i think there's actually a lot of us, but it does feel the convo is lopsided in the other direction (or just not happening, in a lot of places.) so i really appreciate you using your voice for this, especially since it's a harder one for you... at the very least, it's cathartic to hear. i wish i had a solution to offer, but i'll just say: you aren't alone.

[@d.a.n.]

the first slide in the first video was verbatim the ai slop from the first website - so even if it was an person and just the description was AI generated, the content was a likely just copypasta from the ai slop sites

[@SadeN_0]

we're in a tremendous hurry to undermine and undercut human interaction, communication, creativity, and humanity itself like there's something else that matters more? if you're a VC tech bro or web bullshit peddler creating or utilising these things, ask yourself _what the hell are you doing_

[@armokgodofblood2504]

Have you seen the cost of rent now? If some AI company offered me a fat salary to do damn near anything my only question would be "When do I start?"

[@raghvendrapratapsingh690]

One of the MANY downsides of AI is the tsunami of brain rot short form content. So, when a video like this comes along, with no fancy editing, just a person sitting in front of their mic and camera and speaking for an hour on a topic which I care about, holds my attention. I just watched an hour long video where a person speaks on a topic they're passionate about. Thank You for this video and since I'm not as nice as you are, FUCK A.I.

[@bodardr]

I miss the feeling of not doubting of an artist's implication in its artwork. I've developed AI paranoia on the Internet and I don't think it's going to go away. Thanks for nothing, AI.

[@MeriaDuck]

Very true. Every image and text is second guessing.

[@coffeeapple1945]

it's so tragic. i can't engage with anything creative anymore without first scanning for signs of ai slop. even most of the ads i get are for ai products. i think my new year's resolution will be to spend more time outside rolling around in grass

[@goodlookinouthomie1757]

Back in the early 2000s I used to buy a magazine called Computer Arts which featured creative work each month inside and on the cover disc. I used to marvel at the skill and creativity of these artists. Today I'm just mentally saturated with it and sadly I feel very little emotion. In no small part because you just have to assume anything you see is AI generated now.

[@knufyeinundzwanzig2004]

the good side of it is you double check if you really wanna support some random person by buying their products, because it's probably low effort china garbage anyways. so you spend less money and miss out on all of the toxic materials too

[@coreynoodles6876]

Hey Freya, my friend do lighting stuff for video games and recommend this video to me bc it made him feel emotional, I don't do anything close to your content, but thank you for being authentic and sharing your feelings and love for humanity and the story we tell, I certainly felt that love from the other side of the screen

[@nonsquarepixels]

Let us not overlook the sheer comedy of describing an ISO data format specified in 1994 as a "Total Game Changer(tm) for anyone working with 3d designs" 1:12:25 . It almost makes me want to start a podcast where I talk nonsense about "Big Deal" file formats like .BMP and .AIFF for 40 minutes. Working Title: "Is this a virus?"

[@photomatto]

The last minute of your video reminded me so much of the E. M. Forster short story “The Machine Stops” “The story describes a world in which most of the human population has lost the ability to live on the surface of the Earth. Each individual now lives in isolation below ground in a standard room, with all bodily and spiritual needs met by the omnipotent, global Machine.” Plot summary from Wikipedia. Great video. Thank you for making it.

[@jorgeaura2890]

You know, I know I'm pretty much dead inside but hearing and watching people like you that actually care about stuff brings me peace. At least someone cares in this wasteland. Hopefully one day I will too.

[@diribigal]

As an AI-generated GBL file, I also don't like it when people lie. Very relatable!

[@dokxid]

thanks for making us feel less alone, I started running away from all this ai noise and unfounded excitement about it, but I never really had the chance to formulate and think about my thoughts and fears; lovely video, tysm for releasing this

[@dascandy]

Dang, that feeling up to 7:30 ... I was trying to find the specs for a file format (wkb) which has a name that's so useless for searching, and it turned out to be so hard to get through the slop it's quicker to reverse engineer it than to search for a spec and read that. And thanks for the clip at 10:27, that's exactly what my brain does too when it stumbles on a typo or grammatical miss like that. Colleagues ask me why I can't just read across those things and I'm like, no, my sentence reading logic just stops and halts until it figures out what the heck this text was supposed to be. 23:59 the word "to" isn't actually there either. Should somebody make an ASMR recording of broken text like this being read out exactly as written for people to cringe to?

[@JamEngulfer]

Something that makes me think a lot of this *isn’t* AI generated is that when I checked what ChatGPT had to say for “write an explanation of the GLB file format”, the answer it gave was actually good. It had none of the empty repetitive sentences or meaningless words and had more information than all of these sites put together. “Write an explanation about converting GLB to OBJ” yielded good results too, with a step by step guide to doing it in Blender and sample usage for a command line tool. I think the key difference is that ChatGPT actually knows what a GLB file is, while the people writing this SEO slop have no clue beyond a few simple points. Now maybe they’re using AI to assist in the writing process, but they have to be using it pretty wrong to get output at useless as this.

[@paulbonsma9395]

It is AI generated, but the prompt was not “Write an explanation about converting GLB to OBJ”. Giving clear, useful information was not even the goal of these websites.

[@JamEngulfer]

@ Maybe the newer models are too good, but I tried and I can’t get it to generate content that bad. I still think it’s coming from content mills because you have to put in extra effort to remove useful information from the AI output and there’s no reason to do that.

[@EYErisGames]

The first site sounds more like it's translated from a foreign language such as Chinese.

[@sagepolyester4248]

​@@JamEngulfer when generating content at this scale it's very likely not chatgpt but a raw GPT model without the chat-bot presentation that chat GPT has. It is also likely that whoever generated the text is using a smaller model that is cheaper and faster to run like one of the open source models. That doesn't entirely explain the spelling mistakes, ive rarely seen LLMs make typos even in small models. Not sure what is going on there

[@zach9169]

there are websites that use AI to detect whether a string of text is AI or not. these are used by teachers to detect cheating for instance. you can grab a few paragraphs and try for yourself. I'm very confident that most of it will be AI. and if someone thinks "it's just a content mill, AI wouldn't produce this" - sure it can. Just use content mill data as the training data for an AI that is optimized for SEO content. that is likely what is happening imo

[@defnotanny]

AI Spam is replacing actual information and its so sad. If you think its bad now, wait 10 years when 90% of all "users" on the internet will be some AI models. (Social media is already working on making AI users) We need to regulate this, but I think no matter what we do the internet is cooked.

[@esr1412]

I remember searching for some stuff for a bot I was making for a silly event on a chat server. One of the articles had really strange information about some of the features of the app that I didn't remember ever seeing. Then it hit me: it was AI nonsense. The information was extremely wrong. I never trusted anything online ever again.

[@jan_harald]

you mean people would do that? go on the internet just to tell lies?

[@esr1412]

@jan_harald 😂😂

[@RadeonVega64]

they call that a hallucination iirc

[@minhuang8848]

So you just avoided keyword slop and SEO garbage in the twenty years before? Or what made you particularly trusting in a time when Snopes had to be born because spreading nonsense was an age-old tradition? I don't believe that people who are worried about the prospect of an LLM-infested internet have any real idea of how bad it was for most of its popular run. Even the 'custom' content and outlets have only ever churned out one-liners about the release of some new trailer, everything is ranked ever since people saw the money behind the words - and they did that quickly. If anything, it is still very likely to significantly improve the current experience. It's not like I'm going to stop using forums and social media, because half of the user base will just make up facts on the spot. We can compile information and understand what is what, and even today you can use these models in similar ways, which would explain why stackoverflow died, even if these models (on average) haven't already surpassed the core value proposition of such sites. People suddenly want to research things they were too shy to ask strangers online before, and you can pretty much immediately understand how everything related to education is going to have a big moment. Sure, there will be sloppiness and all that along the way, but it can't possibly be worse than what we're dealing with today. The incessant remixes, the repeated cat puns... let's see where it all leads first. I am more than willing to eat dirt if it means that several billion people get to learn pretty much everything in the foreseeable future, rather than realizing that the only chance they had in life was being born rich. And don't get me started on who it enables in the first place. Who cares about my online sensibilities when I see paralyzed people getting as close as they need to be to turn their own imagination and creativity into something tangible, without some clowns yelling at them that 'it's not their art if they can't put it down using their own limbs' or something dumb like that. Blind people got a massive boost. Same for deaf and hard of hearing audiences, in case anyone somehow didn't realize what a big deal automatic captioning has become in the last five years or so. And again, education. There is no alternative university where a non-AI assisted classroom was ever going to solve the problems we keep piling up (especially in the states), now we got very efficient personal tutors engaging students in just about any way THEY could want. I see plenty of issues, but I just don't see how it is remotely worse (now or in the future) than the absolute travesty we've been dealing with for most of human history. Slop me up, as long as you feed me the beneficial discoveries. Not that it isn't already happening anyway.

[@brechdt]

You are so right!! Before AI I even had problems finding the exact stuff I wanted to know because the sensation articles flood the search results and now with AI there's just an infinit amount of flood

[@HBentley92]

Thank you for making this video. It's kind of sadly comforting to see there are others that share my existential dread with generative stuff. It's mind-boggling how many resources are being pumped into something that ultimately provides no benefit to humanity. No one *wants* to read AI generated blog posts. There isn't an audience for these insane 40 minute podcasts about Step files, it's purely a symptom of the insane advertising "content" ecosystem we have online. I wonder if there's a potential for a browser extension that blacklists sites that use generated content.

[@oleksandrbespalov9713]

I'm afraid the internet is like this is for about 15+ years already. Almost every topic I was trying to research looks like this in google. I'm not even trying anymore, I'm looking in youtube instead. Internet was ruined not by AI, but by SEO (search engine optimization) and copy writer horde, which mass produces those empty articles about everything. AI just automated it.

[@khpconan]

Exactly. Search engines have been trash for a long time, 10 or more years. I have noticed it's somewhat worse now, but it was already unusable before

[@kalusharma2201]

Such an awesome video. I feel exactly the same. I'm a developer and keep seeing the tech bros shouting "if you ain't using AI to do bla bla, you ain't gonna make it", "adapt or get left behind" and I just can't understand how people are excited about it. How can people be excited that so many people will loose their livelihood? Its so crazy to me. I've been depressed, and still am, I see no point in trying to improve my craft cuz "AI I can do it in a second". I also resonate with your point about religion. My parents are religious and whenever I discuss it with them, they seem to be somewhat at peace, they say "Whatever the creator wrote, will happen, no point in overthinking and it'll happen to all of us, we'll be in this together". I don't know where I'm going with this, but, yeah, I wish I had this hope too, cuz for me I _know_ that big corps will be the only survivors, and these tech bros saying "AI or YAGMI" will survive for a bit longer, but, ultimately they'll see how cutthroat everything will be and NO ONE will make it (expect ppl at the top). Idk what to say, I honestly want this to be over.

[@Le_Petit_Lapin]

I deeply miss all the old forums from the late 90's to say mid 2010's that have now largely been supplanted by Reddit, Discord and these awful AI generated, SEO optimised blogs. You could always find someone to help you on a specific problem that you had, a human that you could get a reply from and ask further queries back and forth with, but these days they're all dying/dead or just wastelands with no active users.

[@DragonchildTV]

OMG i just saw your video description xD 10/10. And i agree with your points... it used to be "oh, let me google that real quick", now i habe to spend 15 minutes clicking through those endless scroll AI articles just for it to not answer my question... and then compare 5 of them and search in 2 other languages to hopefully get a somewhat reliable answer. Insanity.

[@seowebua]

here is a tip - don't click on domains, that would be unhelpful, like all those io domains in the video obviously would be for some paid platforms. Try to find smaller blogs, that are dedicated to the niche

[@oktaycomu]

Seriously thank you for this video Freya. I’ve been terrified of how AI is getting blindly adopted by many folks around me, and it’s been really hard to explain all the problems we see with it. At least now I can share your video as a quick “Here’s the problem.”

[@ConcerninglyWiseAlligator]

1:06:47 it will get better. AI is not cheap enough to be profitable. Open AI is bleeding money and they will run out of it eventually. People making AI slop have given up on it because it isn't worth it for them to do it. Most of them got into it because it was a get rich quick scheme they were sold on by grifters selling online courses. AI is simply not as good as us. Even though an astounding amount of consumers will just consume uncritically, eventually they'll notice that the quality simply isn't there. This was simply the latest trap room temperature IQ VCs fell into. But when the numbers show that this whole thing is worthless, they will jump ship.

[@amoose136]

I was going mad the other day when I was trying to trace the history of radiation sterilized meat (something that as best I can tell was only a thing for about a year beginning in 2002 at Wegmans and only as ground beef). All the relevant things happened way back then but various search listings were lying about their publication dates to seem more relevant and I also found a few strange articles I do suspect were made by LLMs including a listing on DoorDash that acted like you could buy the thing today but it’s just out of stock even though it hasn’t been available for over 20 years. The search information space is so impossibly bad and even retrieval of old information is considerably degraded.

[@tr1x_trl]

I can't adequately describe the feeling of aversion or maybe even hatred against all of this generated shit, be that text, images or video. I told someone about a project I want to work on, which would involve writing a custom C compiler to compile the original doom source code into some scuffed assembly language I have yet to design, and she asked me if I couldn't just feed it into one of the chat bots. LLMs and all the other generators were never meant to go past curiosities, and I just have to add an increasing number of ublock rules to remove any mention of it from various sites. It was fun when it was just talktotransformer, just gpt2.... spitting out nearly coherent things while sounding like a room of fan fiction writers passing a piece of paper around while absolutely wasted and on the verge of carbon monoxide poisoning. We've just got to continue fighting the enshittification as much as we can... And wow is it incredibly refreshing finally hearing someone else be just as done with what all this has done to the internet

[@tr1x_trl]

Increasingly the only way of getting any real information is from reddit threads and forums that went silent a decade ago

[@tr1x_trl]

I keep feeling like I have things to add The last time generated images still felt like actual art to me was deepdream. Insane hallucinations that just fall out of the math, weird colors, everything turning into eyes, random dog noses from any shape in the image. It was interesting, it wasn't trying to replace humans, it was people making something and fucking about with it

[@kefirahhla]

Thank you so much for this video! I've been telling my friends similar thoughts, but felt alone in my opinions. A few weeks ago, I finished an essay on philosophy of art and AI as a culmination of capitalistic production, and I really enjoy how your ideas in this video complement my thought process.

[@QuickNETTech]

I just checked your channel yesterday to make sure I hadn't missed an upload since it'd been a while, and now I see this beauty today owo can't wait!

[@DustyMusician]

59:00 Nope, this is not true, and I absolutely despise this argument! People say this as if 19th century cameras were disposable Kodaks you can get at the convenience store. They were not! Early photography wasn't useful enough to supplant painting as record of reality due to long exposure times (and the toxic chemicals needed to develop the image), and in fact modern artists had almost immediately jumped to use photography specifically to make more modern art! On top of that, the Impressionists weren't primarily motivated by form or being replaced by the camera but rather subject matter, as in, they were more interested in painting landscapes outside than painting royalty from within a studio. All of the modern art movements who came after them were challenging the form of painting to further the pursuit of painting, not because an outside technology had forced them to change course. This is most evidenced by the fact that portrait painters still exist to this day and haven't gone extinct because painting isn't a purely functional exercise that was "obsoleted" by cameras. Like huh??? I hate that argument so much.

[@DeviRuto]

This doesn't really contradict with the argument, though. It's not the high arts that were replaced with photographs, it was the encyclopedia illustrations, the mugshots, the advertisements. The parts of the industry that don't care about artistic authenticity. We won't get AI bestseller authors, we will get AI listicles and robot customer support. It's also not like the photograph disruption of the visual art industry happened overnight, so the first cameras being shitty is not a good counterargument.

[@DustyMusician]

@@DeviRuto Photography did not "disrupt the visual arts industry." It is frankly bizarre to frame it as such because there was no "industry" to disrupt-art predates industry and is created regardless of it. In any case, painters and other artmakers largely did not and do not see photography as a threat to or replacement for their art. First off, photography is a kind of visual art. No photograph is a perfect record of reality, but you can use techniques such as long exposure to alter the kind of images you can create. So if you want to run with the idea that photographers were "disrupting" painters, you would have to argue that photography's realism was a limiting factor pushing painters to make abstract art, and not only was it not that, but its imperfections became a tool of abstract art almost as soon as it became available to artists. Marcel Duchamp, before he photographed the urinal, was a painter. He switched to photography (technically, he was a sculptor) to make *more art.* Now, in light of my comment, you've limited the scope to commercial illustration; now magically I'm not disproving the point even though the photography argument Holmér uses as an example is explicitly talking about the avant-garde. Don't derail. But on the topic of commercial illustration, commercial illustration is not beyond, in conflict with, or separate from fine art. Alphonse Mucha, Norman Rockwell, and J.C Leyendecker were all commercial illustrators whose influence can be felt among contemporary artists, and their work wasn't displaced by photography, either. Additionally, things like scientific diagrams still require illustrators as photography doesn't or can't provide adequate detail. And again even in spaces where photography does suffice, like botany, we still have illustrators. Concept art and storyboards could be made through collage or photobashing and yet we still have illustrators. Once again, the idea of photography displacing all these painters and illustrators is easily disproved by the fact that there are painters and illustrators who exist today in droves, working just fine. The only way this argument would hold is if it were true that art is purely functional and which kind is used is a matter of efficiency, which is not true, or no one would paint with oils ever again and everyone would switch to gouache.

[@DeviRuto]

@@DustyMusician I feel like you're missing the point. It's not about photography destroying art, it's about the size of each respective market, and the incentives to practice a specific skillset based on the amount of available jobs. It's an economics issue, not one about the definition of what counts as art or not. No one is saying photography isn't art. As a point on efficiency: nowadays movie backgrounds are largely CGI, because it is simply cheaper and faster to produce than hand-painted backdrops. This doesn't mean hand-painted backdrops have disappeared, only that there are fewer jobs that require them, and therefore less incentive to practice that skill. This is what disruption means. Horses still exist, despite the invention of automobiles; it doesn't mean automobiles haven't disrupted the horse industry. And art doesn't have to be purely functional for the argument to hold water. It's not an all or nothing thing; different clients care about artistic integrity by differing amounts. This is obvious to anyone who's seen mobile game ads. Honestly, your "yet we still have illustrators" point sounds a lot like the anti-evolutionists' "why do we still have monkeys?". No nuance.

[@DustyMusician]

@ There are no "art markets" to be discussed 💀 you're misunderstanding the point of comparison when people talk about "AI" and photography. In the video, Holmér paraphrases a common argument people use to support AI images flooding the Internet, which is that it supposedly will help artists be more creative the same way photography did in the late 19th century. The implication is that photography pushed modern artists out of realistic painting and into more abstract art, which is a myth. That's it, that's the topic. You can talk about jobs and markets all you want but that's beyond the scope of the discussion. The point is that photography was not the driving force behind modernist abstract art movements, so whatever point being made about AI in this regard is a flawed premise to begin with.

[@DeviRuto]

@@DustyMusician Disruption is about economics, not artistic movements. The argument was NEVER about photography causing the cultural shift in popularity from illustration to photography (and to cinema, and to everything that came after). That doesn't make sense as being the theme in this context, because, and I agree, that isn't necessarily a problem. The theme of the video and problem at hand is *industry* disruption. People losing access to jobs that use the skills they honed their whole lives. You are the only one here talking about art in the philosophical sense. In the context of the video, art is a JOB. Also, there isn't a market for artists? Most artists are commercial artists and sell their skills to survive! If you get to choose freely what style and field you're specializing in, you're privileged.