What defines AI slop

[FireBorn]

What makes something AI slop? Do you have an objective metric for it? Please explain.

Also, why is 'AI slop' bad, yet Agrippa worshipped? Both collated others work, yet Agrippa gets a pass? Smells funny.

 

[basilicacrafts]

What makes something AI slop? Do you have an objective metric for it? Please explain.

Also, why is 'AI slop' bad, yet Agrippa worshipped? Both collated others work, yet Agrippa gets a pass? Smells funny.

I want to answer this for myself and hopefully do a little education at the same time. All links provided are meant to be jumping off points and also a bit of proof I am not just making things up.
Most people who view AI as slop do so because of the environmental and moral issues with Generative AI. I am not going to get into other types of AI because my knowledge and personal qualms are regarding Generative AI. I also will be just calling Gen AI "AI" for brevity.

AI requires large data centers to run; these

Please, Log in or Register to view URLs content!

in a way that not only takes water away from people in the surrounding areas at an alarming rate and and have been linked to high carbon emissions because of how heavily reliant on energy a gen ai. Anecdotally, I have seen videos and reports of people who live near these centers saying that their water bills have skyrocketed and their water pressure has decreased to a trickle. Considering the ongoing conversations about climate change being primarily carbon emission caused and that just were warned by the

Please, Log in or Register to view URLs content!

on our hands (according to their scientists we have approximately 14 years left of fresh water) it is easy to say that, unless AI changes it's water usage and carbon emissions, it doesn't have a place in a future where climate change improves and the human race continues living without the suffering caused by lack of resources.

AI also has moral issues wrong with it as most of the data collected for image/video/text creation was stolen from creators. There are laws in place in the USA to protect creators from theft, but because AI is new and untested (legally) they simply got away with it. And while there are currently lawsuits that artists/musicians/writers/etc are keeping an eye on (the recent win of anonymous images created with AI cannot be copyrighted comes to mind), they are neither going fast enough nor back paying creators for their stolen content. I cannot fight the exact quote but I remember one of the larger AI CEO's saying something along the lines of they couldn't make AI do what it does without theft - so even if there was a way to create AI content ethically, there wouldn't be enough opt-in to make anything good (not that I think any of it is good).

This is not to even touch on the way AI usage (usually in LLMs)

Please, Log in or Register to view URLs content!

and increased forgetfulness,

Please, Log in or Register to view URLs content!

, is often incorrect with information, are a threat to security, is a tool of fascism, and has made it so I and other AI aware people cannot see anything on the internet without having to distrust that it is real (honestly, I just want real cat videos - why do we need fake cats? The real ones are so much better.). (Some of these are without links, please just do some googling.)

I think supporting a theft-powered machine is bad enough. As an artist I find that, from my most objective standpoint I can muster, the images/writings/etc AI creates are horrible. AI doesn't have the ability and I hope it never does to understand the complexity of good art. It isn't just extra or messed up fingers, it is the lack of understanding composition, contrast, tangents, etc. AI has learned that art pieces require color and a subject but doesn't know how to put it together like an artist does. AI does know that there are beats in a story many people put into it, but it doesn't understand why an author does those things.

AI doesn't understand the less tangible and defined parts of art which is being human.

The machine can "write" and "paint" and "photograph" like a human because it is trained on human works but it doesn't understand they why of it all. The deep human urge to create and enjoy the process and to connect with the gods given gifts we can mold into something more and more powerful and connective. The 1979 IBM Training Manual says: "A computer can never be held accountable, therefore a computer must never make a management decision." Why? Because human nuance is vital to the expression of what humans think is right and wrong. That goes for the macro to the micro: managing employees to managing what line goes where.

So to more directly answer the question: AI is determined to be slop because it meets the definition: waste that often has to be removed by hand; a byproduct with little to no value.

As a human, Agrippa may have created collages of thoughts and collations of others' works but did so as a human and with his own input. Agrippa didn't use 5 million gallons worth of water a day to do so either. Agrippa created a compendium of human knowledge rather than mashing it all together in an incoherent mass of text. Even the mistakes in compiling those ideas are human in nature.

 

[FireBorn]

I want to answer this for myself and hopefully do a little education at the same time. All links provided are meant to be jumping off points and also a bit of proof I am not just making things up.
Most people who view AI as slop do so because of the environmental and moral issues with Generative AI. I am not going to get into other types of AI because my knowledge and personal qualms are regarding Generative AI. I also will be just calling Gen AI "AI" for brevity.

AI requires large data centers to run; these

Please, Log in or Register to view URLs content!

in a way that not only takes water away from people in the surrounding areas at an alarming rate and and have been linked to high carbon emissions because of how heavily reliant on energy a gen ai. Anecdotally, I have seen videos and reports of people who live near these centers saying that their water bills have skyrocketed and their water pressure has decreased to a trickle. Considering the ongoing conversations about climate change being primarily carbon emission caused and that just were warned by the

Please, Log in or Register to view URLs content!

on our hands (according to their scientists we have approximately 14 years left of fresh water) it is easy to say that, unless AI changes it's water usage and carbon emissions, it doesn't have a place in a future where climate change improves and the human race continues living without the suffering caused by lack of resources.

AI also has moral issues wrong with it as most of the data collected for image/video/text creation was stolen from creators. There are laws in place in the USA to protect creators from theft, but because AI is new and untested (legally) they simply got away with it. And while there are currently lawsuits that artists/musicians/writers/etc are keeping an eye on (the recent win of anonymous images created with AI cannot be copyrighted comes to mind), they are neither going fast enough nor back paying creators for their stolen content. I cannot fight the exact quote but I remember one of the larger AI CEO's saying something along the lines of they couldn't make AI do what it does without theft - so even if there was a way to create AI content ethically, there wouldn't be enough opt-in to make anything good (not that I think any of it is good).

This is not to even touch on the way AI usage (usually in LLMs)

Please, Log in or Register to view URLs content!

and increased forgetfulness,

Please, Log in or Register to view URLs content!

, is often incorrect with information, are a threat to security, is a tool of fascism, and has made it so I and other AI aware people cannot see anything on the internet without having to distrust that it is real (honestly, I just want real cat videos - why do we need fake cats? The real ones are so much better.). (Some of these are without links, please just do some googling.)

I think supporting a theft-powered machine is bad enough. As an artist I find that, from my most objective standpoint I can muster, the images/writings/etc AI creates are horrible. AI doesn't have the ability and I hope it never does to understand the complexity of good art. It isn't just extra or messed up fingers, it is the lack of understanding composition, contrast, tangents, etc. AI has learned that art pieces require color and a subject but doesn't know how to put it together like an artist does. AI does know that there are beats in a story many people put into it, but it doesn't understand why an author does those things.

AI doesn't understand the less tangible and defined parts of art which is being human.

The machine can "write" and "paint" and "photograph" like a human because it is trained on human works but it doesn't understand they why of it all. The deep human urge to create and enjoy the process and to connect with the gods given gifts we can mold into something more and more powerful and connective. The 1979 IBM Training Manual says: "A computer can never be held accountable, therefore a computer must never make a management decision." Why? Because human nuance is vital to the expression of what humans think is right and wrong. That goes for the macro to the micro: managing employees to managing what line goes where.

So to more directly answer the question: AI is determined to be slop because it meets the definition: waste that often has to be removed by hand; a byproduct with little to no value.

As a human, Agrippa may have created collages of thoughts and collations of others' works but did so as a human and with his own input. Agrippa didn't use 5 million gallons worth of water a day to do so either. Agrippa created a compendium of human knowledge rather than mashing it all together in an incoherent mass of text. Even the mistakes in compiling those ideas are human in nature.

I think part of the issue here is that several different criticisms are getting mixed together under the label “AI slop,” and clarity matters.

If someone means they object to AI on environmental or moral grounds, that is one conversation. If they mean the books are low-quality, derivative, inaccurate, or padded nonsense, that is a different conversation. Those two things can overlap, but they are not automatically the same claim. Again, clarity matters.

The reason I asked what makes something “AI slop” is because I do not think the label means much unless we define it. Is it slop because AI assisted in the process at all? Is it slop because the author published a lot in a short time? Is it slop because the writing is shallow, repetitive, or factually weak? Those are very different standards.

Occult literature has always involved compilation, translation, borrowing, reframing, and commentary on earlier material. That is not new. So if the criticism is simply that an author is collating existing ideas, that alone does not prove anything. The better question is whether the material shows discernment, structure, and understanding, or whether it just reads like plausible-sounding text with no depth behind it.

I can absolutely see a case where someone with no grounding in a subject asks AI to write a book on Witchcraft and publishes the result full of errors. That would deserve criticism. But I can also see cases where someone knowledgeable uses AI as an assistant for organization, comparison, or drafting, and the final work is still solid because the human author actually knows the territory.

So to me the issue is less “was AI involved at any point?” and more “does the book hold up?” Is it coherent? Is it accurate? Does it show real understanding? Those seem like more useful metrics than assuming recent publication date plus AI cover automatically equals slop.

I am not saying anyone has to like AI. I am saying we should be careful not to replace discernment with a reflex label. Hope that clear up my point. Not trying to be adversarial, just asking for clarity because I think it deserves some effort in that direction.

 

[PracticingGeomancer]

What makes something AI slop?

If it's text (or images, for that matter) that are generated by a clanker and not by a human. There are nuances to this: what about text that a human wrote and a clanker re-wrote to make more grammatically or orthographically correct? How about text that a clanker generated and a human reviewed? I'm not an expert on the matter, I don't know.

Also, why is 'AI slop' bad, yet Agrippa worshipped? Both collated others work, yet Agrippa gets a pass?

We can assume Agrippa was not a generative LLM. These are two matters conflated into one:

• is Agrippa "worshipped" because his material objectively is high value? Or because he's a very old source -- therefore correct according to some? Or a mix of the two? It's an interesting question, albeit slightly off-topic for this thread
• AI generates text by guessing the most likely next word (brutally oversimplified), it lacks understanding. Humans can also lack understanding, sure, but clankers lack it completely. @FireBorn if I wrote a book using dice to select the next most likely word from a pool of likely next words, would you use it to learn a new skill?

That being said, there are many other issues with clankers, as @basilicacrafts lifted earlier.

I am not saying "any author who uses AI is bad by definition",I am saying an author spewing out several dozens of books in the span of less than 2 years doesn't look too promising. It seems like a low-effort moneygrab. That being said, as I wrote in my post above, I will actually give Andrew a fair change, and read one or two of his books.

 

[SkullTraill]

Off topic posts moved to a new thread.

 

[SkullTraill]

Also, why is 'AI slop' bad, yet Agrippa worshipped?

Effort. I love AI, and I use it often for research, deep searching the web when I don’t know the right search term, looking through documents to find what I need and or summarize etc.

However, think of the effort, skill and dedication it took Agrippa to write his books. Vs someone who can push out 100s of books a month using AI.

Using AI in and of itself does not make a work “AI slop” but using AI alone to generate content typically makes that content AI slop. AI is great for finding things and summarizing things. But when you ask AI to expand on something it typically does not do a great job. It might look good but it’s the equivalent of babbling.

 

[Van Horne]

I'm not at all an expert on Agrippa, so I can't say anything regarding this. I'm not at all hostile against the use of AI, rather open minded towards new technological advancement. I believe that the peak of practical use and in-general potential of AI lies far ahead in the future.

The problem I see at the moment is that people use AI in today's state to compensate their lack of creativity and motivation. The result of this is what I scorn so much. It's not only about the book market, it also affect the music market, and the internet in general. Already before the rise of AI there was a palpable decline in creativity and diversity, through business strategies, design decisions and a gigantic human glob of people with too much time and money at their hands and internet access spamming the digital realm with their generic work, music, art, literature, and so on. And AI just set this gigantic pile of trash on fire.

To me, the most important thing is aesthetics in every aspect. It's hard to grasp and much harder to describe but in the aesthectics of a certain product lays the essence of it's genesis, the afford of human creativity. And AI slop is missing all of that.

An example from the world of occult books:
The first time I saw the three trilogies of Kenneth Grant I knew I had to have them in my collection. I was drawn to them simply be looking at the cover artworks painted by his wife Steffi. It was so weird and strange and in the same way mystical and foreboding. And, oh boy, when I started to read them, the aesthetics did not lie to me. What a glorious madman Grant was! His mind was sprawling with ideas, interconnections and absurd conclusions. This kind of creative insanity you cannot simply reproduce, but this what I enjoy so much.

And now, take a look on the occult book market. Before AI it was already a quite mediocre slop of recycling and regurgitating ideas, with exceptions, of course! "Here, take a look at my new 'grimoire', Celtic Qabbala for the Queer Witch!" And now this horse crap comes with insufferable generic AI artwork, probably not even written by themselves, and the pile of trash is burning with suffocating black smoke.

Sorry for my rant!

 

[FireBorn]

Effort. I love AI, and I use it often for research, deep searching the web when I don’t know the right search term, looking through documents to find what I need and or summarize etc.

However, think of the effort, skill and dedication it took Agrippa to write his books. Vs someone who can push out 100s of books a month using AI.

Using AI in and of itself does not make a work “AI slop” but using AI alone to generate content typically makes that content AI slop. AI is great for finding things and summarizing things. But when you ask AI to expand on something it typically does not do a great job. It might look good but it’s the equivalent of babbling.

Hmm so the assumption of effort for Agrippa to crank out his 3 books in three years vs. AI farting out 100s of titles a month speaks to speed, not effort (apples and oranges). I see what you are going for here, and agree that unguided AI leads to babbling, but it's still messy.

What if an occult author used AI to help write 3 books in 3 years? That is a more accurate comparison, does that change your opinion?

For context Agrippa wrote his 3 books in a three year span (1531-1533), amid his travels and legal troubles, then dipped out of the occult scene altogether. So the effort idea might be more romantic than accurate. He reframed and added commentary to the major topics of his books, he didn't create all of it. Sound familiar? Had he had access to scribes, he might have used them. Would that have made his work slop of that era?

Effort is subjective and might not be a solid metric. Who determines effort? Lots of books are filled with babbling. Again, subjective. This is much harder than it appears on the surface and that is kind of my point here.

Political vibes is opinion. Effort isn't consistently measurable. Coherence is sort of subjective. Accuracy is measurable, but not all books are academic. Back to square one. AI is just a tool, nothing more.

Maybe the cleanest method would be to look at the author/generator of the work. The tool didn't suck, the author/generator using it was lazy/ignorant/negligent. Putting an AI on auto pilot to write a book is definitely a low level thing, but obviously not the totality of the issue here.

This is the harder part, it requires discernment and effort on the readers part in my opinion. The term AI slop is just an opinion, sometimes agreeable, sometimes not.

 

[SkullTraill]

What if an occult author used AI to help write 3 books in 3 years? That is a more accurate comparison, does that change your opinion?

I believe I was already clear about my opinion that using AI is not a problem, and in most cases something I would encourage, but not to generate text. So it depends. Did they use AI to, look up information, structure their writing, fact check, proof read, etc? Or did they use AI to generate text instead of writing it themselves, and if so, did they generate in small chunks with their supervision, or did they not even read what the AI generated?

Effort is subjective and might not be a solid metric. Who determines effort? Lots of books are filled with babbling. Again, subjective. This is much harder than it appears on the surface and that is kind of my point here.

The reader determines it, along with everything else about the book. Literally everything about a book down to the number of pages it has, is subjective. Many people don't look favourably on AI generated works, because quite literally still to this day the most advanced AI LLMs tasked with generating long form informational text (especially on niche topics) inevitably ends up generating a significant percentage of garbage.

Putting an AI on auto pilot to write a book is definitely a low level thing

That's what I'm saying.

There's also something to be said about the market being flooded with slop - and again, to define slop, I mean something generated that is fantastical and an output of a predictive language model with hallucinations that are not real facts and has zero value to a reader - because it makes it harder to find actually good books.

Also a thought experiment: if AI becomes so advanced that someone with 0 occult knowledge, and 0 writing skills can ask a AI to generate "the most useful and informational occult book" and the AI is able to, and does just that. Then yes, surely to book has value. But does the "author"? I would argue not, and that that person is neither an occultist nor an author, but rather a prompter. And if an AI is perfect, then the prompter is worthless.

 

[Ziran]

What makes something AI slop? Do you have an objective metric for it? Please explain.

AI can't do literary analysis. It can't probe into layers of allegory. It's a parrot.

Also, why is 'AI slop' bad, yet Agrippa worshipped? Both collated others work, yet Agrippa gets a pass? Smells funny.

It's a time-suck

 

[KjEno186]

Every human has a unique way of writing, and it can evolve over time with knowledge and practice. Having used a variety of LLMs locally, I know that each base model also has a unique way of writing. There are communities dedicated to creating remixes of open source models to improve writing styles and 'intelligence'.

Anyone who has used LLMs will notice the writing styles, peculiar phrases, and formatting unique to each model. Perhaps, if one could do a homemade remix that no one else in the world had access to, something quite out of the ordinary, it might be possible to pass it off as one's own writing. To my way of thinking, the "slop" becomes apparent in the usage of the same model or variant by millions of people. The practiced eye can see right away that it fits a particular pattern of words and phrases. One begins to suspect that it is machine generated. This is the "uncanny valley" of writing. Humans seem to have a revulsion to things which appear to be, but are not quite, real.

Slop is this Uncanny Valley of the written word. This phenomenon extends to the visual output of AI models intended for images and videos just as it did for older CGI. I think a significant portion of the population has developed a tolerance for machine generation of media because the technology as been around for decades now to mimic 'reality' through artificial means. However, prior to generative AI, the technology was still in the hands of skilled operators. The democratization of the software and availability of powerful computers has put this creative power into the hands of those who churn out volumes of media, and we're able to recognize it for the low effort product that it is.

 

[Morell]

What makes something AI slop? Do you have an objective metric for it? Please explain.

Also, why is 'AI slop' bad, yet Agrippa worshipped? Both collated others work, yet Agrippa gets a pass? Smells funny.

In my opinion it is just rode way to say that something is made by AI, revealing disgust in the fact that people use AI.

 

[beardedeldridge]

First I think the question as posed (regarding Agrippa) is a bit… anyway maybe you’re just trying to goad people into a discussion on AI which is cool.

So if we assume the 3 works didn’t exist and then tomorrow some dude named Agrippa dropped the works out of nowhere and everyone just considered it gospel and ran with it then that could be considered getting a pass.

Now Agrippa did largely just collate others (which is what a lot of practitioners the world over do - at least when they get started) work but the fact that he did it fairly accurately and helped spread this tradition is why it is held in such high regard. We can see this when his works are compared to manuscripts of that era along with earlier and later ones.

You add that to the fact that practicing magicians who used his work directly or used the later works that used his works as a source say it worked, further adds fuel to the fire that he didn’t just get a pass but that over a long period of time the work has been tried and tested and held up fairly well. Now do I consider it the magicians bible that is beyond question, no, but anyone who says it wasn’t an important work that had an impact on the magical community as we know it today - is an idiot along with anyone else who doesn’t question what they are reading.

Now if an AI happened to have existed back then and the manuscripts of the era were somehow digitized and it produced a similar text that got largely ignored then yeah you may have a point.

Now do I find AI worthless in occult research, NO. I often use it when I know I read something a while back and can’t remember where or as a first step in my research into something newish like skull mentioned above.

From another discussion someone asked a little while back, I did end up using one of the widely available AI tools to find and reproduce a detailed and accurate ritual from a grimoire. It only took a few tweaks to the original prompt but it would have been almost as easy to just pick up the grimoire and write out the outline and script myself but it could prove useful if you are having a hard time due to the ritual being broken up and spread throughout the grim which I think was the OP of that threads problem if I remember correctly.

Now in my non-occult work. I’ve been using it and comparing the results against the analyst team’s results. The last results I had with AI (which is inline or better than early results)

Team produces scenarios A, B, C, and D. Now historically 50% of those end up happening.

AI produces scenarios C, D, E…L. So it is coming up with some of the teams scenarios but missing others but from my analysis E-L fall somewhere between unlikely to batshit crazy and the AI results are more around 10%.

All that to say its fine to use AI just check and double check the results.

Now to answer the title question - I consider AI slop (still not a good question) to be when the prompter turned their brain off and just went with whatever the AI told them. That being said theres plenty of human slop out there as well so don’t just blindly follow that either.

-Eld