AI Derangement Syndrome / Anti-AI artists / Pro-AI technocultists / AI "debate" communities - The Natural Retardation in the Artificial Intelligence communities

[SoleniyKraut]

Artists tantruming about AI are shit artists.

I agree, in part. Back in the day, references were either some sketch you found in a book, or a person\object you intentionally observed or placed as a reference. Being a person that artists used as a reference was a job in of itself.

Photos aren't bad as references, technically, but over time people began literally tracing photos, even using the grid method to completely copy them onto paper. I've seen so much people fawn over arist accounts where the guy just redraws a photo one to one. It's cool, perhaps it takes a lot of time. But I have the photo already, why do I need a drawn rendition of it?

 

[AmpleApricots]

Did you ever try weboob?

No, but I'll look at it and maybe even hand them to the agent. Problem with such tools is often that they always eventually break because of API changes etc.. The internet already became very unfriendly to anything automated with cloudflare and friends.

The cool thing about such harnesses like hermes is that you can do things like say "Hey, the pope wrote some AI manifesto a few days ago, can you read it and tell me what it is about and also send it to me as markdown file" and it'll just happen and you don't have to concern yourself with the details like where he will find the manifesto or who even the current pope is. Nothing but the pure text at hand which you can then process and permutate endlessly. I think LLMs here have a unique strength because of their ability to process language and deal with ambiguity. Especially the ambiguity thing is very powerful when used correctly. Before LLMs, that was an exclusively human thing that just couldn't be solved with classical computing approaches in a satisfactory and non-brittle way.

This also means that an LLM/Agent will (in the best case) always be able to adapt to these changes in the web contrary to these static tools that'll need an update that might never come. Who genuinely cares if the scraper he wrote for that government website to make an appointment is a bit shit? Important is that my time is saved and I didn't have to deal with it and I can keep that mess out of my clean world of text strings.

EDIT: To stress the usefulness of permutation and reprocessing and "learning" (in a way): If the model writes poor lisp, you can direct it to read practical common lisp which is an excellent book that's free online and write a markdown-formatted guide for itself that it can use as a guideline to write future lisp projects. If that actually would work well I don't know but it demonstrates the power of these harnesses pretty well IMO.

 

[Hannibalistique]

No, but I'll look at it and maybe even hand them to the agent. Problem with such tools is often that they always eventually break because of API changes etc.. The internet already became very unfriendly to anything automated with cloudflare and friends.

The cool thing about such harnesses like hermes is that you can do things like say "Hey, the pope wrote some AI manifesto a few days ago, can you read it and tell me what it is about and also send it to me as markdown file" and it'll just happen and you don't have to concern yourself with the details like where he will find the manifesto or who even the current pope is. Nothing but the pure text at hand which you can then process and permutate endlessly. I think LLMs here have a unique strength because of their ability to process language and deal with ambiguity. Especially the ambiguity thing is very powerful when used correctly. Before LLMs, that was an exclusively human thing that just couldn't be solved with classical computing approaches in a satisfactory and non-brittle way.

This also means that an LLM/Agent will (in the best case) always be able to adapt to these changes in the web contrary to these static tools that'll need an update that might never come. Who genuinely cares if the scraper he wrote for that government website to make an appointment is a bit shit? Important is that my time is saved and I didn't have to deal with it and I can keep that mess out of my clean world of text strings.

Yeah, I think LLMs will be the final solution (so to speak) for the problem weboob tried to solve.

FWIW both weboob and hermes are written in python, so could potentially integrate it directly somehow instead of the overhead of HTTP/MCP or normal shell tool calling (though I haven't looked at hermes since its public release, so no idea if its feasible).

 

[Sheogorath]

i wanted to read about techno cultists thinking autocomplete will tell them the truth about the dimensional merge but it's all coping that having a robot draw pictures for you makes you an artist

 

[Hannibalistique]

i wanted to read about techno cultists thinking autocomplete will tell them the truth about the dimensional merge but it's all coping that having a robot draw pictures for you makes you an artist

Be the change you want to see in the world.
Go to reddit and find some of the esoteric LLM subreddits, and harvest screenshot for the thread.
(Personally I got bored of them and/or don't have the cogsec to wade through reddit to find actual human posts about spiralism or whatever.)

 

[femboy fart huffer]

Personally I got bored of them

Almost all of it devolved into unentertaining bullshit. It's either retards parroting the same reddit misinformation talking points or grifters hyping up chatgpt because 6.0 will totally be sentient you guys totally please invest in the company so I can dump my shares on you haha lol teehee.

 

[harmy.kook]

but

You know it's funny you say that while there's a post at the top of the page pointing out about the two main groups buying into the overhyped, but one being extra faggy.

Have you considered that one side is allowed to be way more vocally and publicly deranged? As such, being more notable. Especially seeing how even being neutral is enough to get you chased out?

 

[Battlefield42]

.

I'm making claude reverse engineer a 3d model format. I'm not sure if it's going to work or not.
Update: It worked, model shows up fine in Blender. There are issues though with vertex groups. Going to sleep now so i can test more tomorrow after work.

 

[pollution jim]

Welcome to the Information Age, the Information Economy, the Information Rwvolution. Expect Rapid Onset Ludditism to explode as those human animals specialised to thrive in The Age Of Mechanical Reproduction continue to find their old tactics impotent.

Consider the art of pedagogy, i.e. teaching. Infamous for strikes and workers' actions, these, The Real Heroes, have grown fat holding hostage the intellectual development of the generations entrusted to them. Today, they go on strike to prove how important their Elite Human Kapital is, and whoopsies, they instead prove that Gemini is a better teacher, a cheaper teacher, who doesn't show up late or sacrifice your children for its entitlements.

This isn't the XIX century anymore. This isn't the XX century anymore. Labour is not the Wunderwaffen it once was. Strikes don't work in the Informational Mode of Production.

please take your meds

 

[ADHD Mate]

Think about it, if AI meant a worker could do 5 times the work they used to, I could keep my workers and multiply my effective work done by 5 for presumably little cost. So my operating costs would remain similar, but stuff done would increase. If I instead fire workers so I am producing the same amount of output as I was before using less, my revenue in theory should increase, but any competitor who kept employees on will be out-producing me. Which is better long term?

The issue is that you assume there is infinite amount of meaningful work to be done. Which is not the case, especially in similar problem space.

Simple thought experiment:
If you have shoemakers that each cam make 1 pair a day, and then you introduce machine that will make 10 pairs a day with 1 shoemaker.
You won't have market to sink 10x more shoes. Even if the price goes down to 1/10, people wont just buy 10x more shoes.
What's worse you competition might now have increase their production as well, so you will have harder time competing with each other.

There is a reason why industries go through boom and then merger and shrinking cycles. All optimization and automation leads to it.
By the way I am still skeptical of efficacy of AI. Just pointing out that idea of increased output being always better long term decision is wishful thinking.
We are already at a point in tech where people complain that programs/platforms try to do too many things for no reason.

 

[Greedy Fly]

i wanted to read about techno cultists thinking autocomplete will tell them the truth about the dimensional merge

May I recommend the Yudkowsky thread or the Zizian thread?

 

[Waifuwars]

Photos aren't bad as references, technically, but over time people began literally tracing photos, even using the grid method to completely copy them onto paper. I've seen so much people fawn over arist accounts where the guy just redraws a photo one to one. It's cool, perhaps it takes a lot of time. But I have the photo already, why do I need a drawn rendition of it?

Relying on photos is not great since a camera sees things differently. There's a reason why the art schools require students to do live figure drawing and to actually touch grass and go outside to paint/draw as it also trains your eye and helps you understand motion, light/shadows, etc. They throw a ton of shade on digital art since they rightfully see it as a crutch and as a far more commercial product.

Almost all of it devolved into unentertaining bullshit. It's either retards parroting the same reddit misinformation talking points or grifters hyping up chatgpt because 6.0 will totally be sentient you guys totally please invest in the company so I can dump my shares on you haha lol teehee.

Josh's recent AI art segment pulled some AIDS comments but I swear these 'people' have to be bots with the way they keep spewing the same comments. "My water and electricity!" "SLOP SLOP!!" We need some next level schizo posting at this rate

 

[The Ultimate Ramotith]

True art:

https://youtube.com/watch?v=YGyvLlPad8Q

…and I got an ad of the AI-motivated Sword and Staff mobile game.

I found this:

TLDR: because firing people would ge these corporations into legal trouble, they replaced them with AI, fully knowing that AI s not 'here yet', then publicly ditch the AI and either fill the gap with H-1B enefactors or have no replacements.

Boing back to my previous post, hough I doub th is intentional, I wonder: if bullying animation hierophants off using AI would result in they being replaced by people who do not care (that much) abou the craft of animation, the result would be AI-generated garbage. That would essentially create 'proof' that anythin generated by AI will invariably be garbage. Considering who receives H-1B enefits, this may even be a 'woke' way of causing 'Asian hate'.

 

[Disruptive Poster]

0:16

 

[Colon capital V]

Josh's recent AI art segment pulled some AIDS comments but I swear these 'people' have to be bots with the way they keep spewing the same comments. "My water and electricity!" "SLOP SLOP!!" We need some next level schizo posting at this rate

Yeah kind of weird some comments act shock that Josh, the forum admin/developer who's blacklisted in every conceivable way financially, likes to use AI for his workflow.

 

[femboy fart huffer]

likes to use AI for his workflow.

Josh actually loves ai so much the only reason he runs this site is to accumulate as many images and videos of fat women to train an ai on them. There's actually a secret board where all of the deathfats have been ai video'd into eating obscene amounts of cheese and burping really loud.

 

[Orange Rhymer]

One thing I love AI for:
Ending the 'Green' bullshit.
Kinda funny. When Meta/X/Google decide that they need to burn more fuel in a year, than last centuries combined human output. Green Activism just...died.

Weird.

 

[Hannibalistique]

One thing I love AI for:
Ending the 'Green' bullshit.
Kinda funny. When Meta/X/Google decide that they need to burn more fuel in a year, than last centuries combined human output. Green Activism just...died.

Weird.

Yeah, Google have been pushing their "green" data centers for as long as I can remember, but they suddenly started being very quiet about it right around the time they released PaLM iirc.

 

[Runch]

Eh, a few months ago Microsoft introduced an all new data center design that's fully closed loop and only needs to be flushed and refilled every 6 years or something like that. I think it's called Fairwater?

 

[Hannibalistique]

Eh, a few months ago Microsoft introduced an all new data center design that's fully closed loop and only needs to be flushed and refilled every 6 years or something like that. I think it's called Fairwater?

They all still have sustainability departments and whatnot, but e. g. Google quietly removed its prominent net-zero-by-2030 branding on its sustainability page in 2025.
More efficient data centers are generally cheaper to run (and easier to set up/run), but the green-at-all-costs rhetoric seems to have mostly died in public messaging.