Updated: May 16, 2023
Witch books are flying off the shelves, especially on Amazon. The problem is a vast number of them, possibly a majority, are fake AF. Written by A.I. or ghostwriters contracted on places like Fiverr from people with zero previous knowledge or experience in the art & practice of Witchcraft, it's almost no surprise we have a whole generation of Witches who feel lost and confused. These books are badly written, full of disinformation, and they take business away from actual authors and practitioners who are out here doing the fucking work.
Books are certainly one of the biggest ones, but you can also find A.I. writing blog posts, social media posts, articles and more. It's everywhere. I get bombarded daily on my social media feeds with invitations to sign up for A.I. content creator sites. Clearly there are folks looking to make their life easier who that appeals to. But the cost is too high when the end result is loss of quality and the dumbing down and watering down of information. Real authors can not compete with this nonsense. When you have to sit down and actually write out your own content, it takes time. You can't just churn out dozens or hundreds of books, blog posts or other content in a matter of days or weeks. But folks who use A.I. can.
Tangent: To be clear, I'm against all badly written books and content on Witchcraft, not just A.I. generated ones. I don't think you have to be the best writer, either, just know what the fuck you are talking about. That's all I ask. I'm over every person who reads one book or does a handful of spells deciding they are now an expert and should teach or write about it. I want back the days when someone spent years, often decades, mastering a craft and then went into teaching or writing. Where are those people?? But I digress.
I've talked about this multiple times on my own platforms and will keep trying to raise awareness around this issue for as long as it continues to be a problem. Previously I focused more on ghostwriting, so in this blog post I will focus more on the A.I. side of this issue.
You can help by sharing this blog post, talking with your friends or inviting a discussion around this topic that prioritizes facts and resources with your covens, groups, and community members.
Keep in mind that A.I. programs are evolving fast. While they can't quite pass a Turing test just yet, they may be able to one day. The information I offer below is the best advice I can give you based on what A.I. can do today. A.I. content creation programs will continue to get better, and they will become harder to distinguish from human-generated content.
"AI today is surprisingly good at producing very fluent, very grammatical text. But it does make mistakes. ... Machines make distinctive types of errors -- common-sense errors, relevance errors, reasoning errors and logical errors, for example -- that we can learn how to spot." -Liam Dugan, Ph.D. student at Department of Computer and Information Science
So let's get into it.
1. LENGTH OF SENTENCES
Look for very short sentences. A.I. generated content isn't very convincing when it comes to consistent long-form content. It tries to mimic human writing but it's just not very good with extensive sentence complexity. Yet. Most humans write with a few short sentences mixed in with more complex and interesting sentences. Since A.I. hasn't really figured this out yet, it becomes clear to experienced readers when you get into it that this writing rhythm and complexity are missing.
2. REPETITION
A.I. likes to talk in circles. I sometimes repeat myself and talk in circles, too. That's because I'm old and my brain is tired. But A.I. repeats words and phrases because it doesn't actually know what it's talking about. Since it can't speak on the subject matter well, it tries to gloss over and cover up its gap in knowledge like a high school student who writes extra large and double spaces every letter to meet the page minimum. The A.I. does this by trying to fill up space with relevant keywords that it's been tasked with, which results in a lot of repetition.
3. LACK OF CREATIVITY AND SELF-REFLECTION
When people write a book about Witchcraft, they are drawing on their experiences that include mistakes and failures, successes and lessons learned. This requires some self-reflection and complex analysis on the part of the author. As a machine, A.I. is good at collecting data, but not so good at turning that data into something meaningful. Content that feels very cookie-cutter or like it's just a list of facts and lacks insight is a big red flag. When a real human experiences something and reflects on it they are able to talk about and go deeper into that idea, technique or experience in a way A.I. (so far) can not.
A.I. also can't really do anything unique because it's just recyclings bits and bobs from existing content. Look for that uniqueness and creativity to help you navigate away from A.I.
4. MISINFORMATION & BAD DATA
This one infuriates me. I hate books on Witchcraft that give new folks bad information. A.I. contributes to misinformation by collecting data from multiple sources, regardless if they are good (accurate) sources or not. So mistakes happen. A.I. systems are often tasked to learn from low-quality fake news. Then they use that bad information to generate even more believable content, resulting in worsening misinformation.
A.I. is like that bad friend who repeats every rumor they hear without trying to find out if it's even true or not. Girl, bye.
Sometimes you can spot these discrepancies when the same 'facts' are repeated but don't match up. That's because different sources online might say different things, and A.I. doesn't have an opinion or the ability to notice that it's getting conflicting information. It just repeats what it hears.
The meaning of words can vary over time and this brings up another troubling side-effect of A.I.'s inability to do this well. It can result in the A.I. including prejudiced language based on gender, race, or geography because it doesn't understand those issues or nuances. It can also use language that in different contexts can be dog whistles based on which political party the reader belongs to. This can even worsen societal prejudices when shared and read by a lot of people, such as what we've seen happen on social media platforms.
5. KNOW YOUR AUTHORS
This one is similar to what I suggest for ghostwriting. You have to kind of be your own detective if you really want to support human authors and avoid A.I. content. Remember that A.I. generated content is found in blogs, social media posts, e-books and more. You would probably be shocked by how many of these you've read or followed that are actually A.I.
Start by establishing the credibility of the author. Visit their author's page and find them on social media. It's completely valid for someone to not be on social media or have an author's page, too. What I look for regarding red flags is does this person have some involvement in the community they are writing for? Have they done book signings? Have they attended any festivals or cons? Have they done any classes, speaking engagements or events? Can you write to them or contact them? If so, what do you get back? Are there people who have met them, and are they sharing those experiences or stories in groups or online? Are there pictures of them with fans, customers, publishers, or recognizable names in the community? There may be plausible explanations for any one of these to not exist, but when too many of these can't be verified, that starts piling up and that's too many red flags to ignore, imo.
6. VERIFY SOURCES
If they exist, actually click on the links and sources in articles, blog posts, social media and even e-books. Do they lead somewhere or are they just for show? Do they lead somewhere credible, or to more questionable shit? Do they lead right back to the author's page or book even when it was supposed to take you somewhere else, like a study or data to back up a claim? Sometimes writers are lazy, don't do their research or rely on bad data, too. It could also be an indicator of A.I. writing, so stay diligent. A.I. doesn't use the library for research, so if the Bibliography or 'Resources and References' section (if it even has one) includes a bunch of basic websites like ehow.com, tarot.com, nasa.com and the like, be suspicious.
7. ILLUSTRATION AND PHOTO CREDITS MISSING
There are plenty of sites that fail to give credit to artists and source content because they are lazy, but if you are reading an actual book or e-book and marveling over the art, illustrations or photos, check those credits. It's almost unheard of for A.I. generated books and content to credit sources properly.
If credits appear to be there but you are still getting hinky A.I. vibes, put on your detective cap and follow the breadcrumbs. Questions you might ask... Did those artists actually create those images, meaning, are the right artists credited with the work? Also, did those artists give permission for their content to be used this way? That may not be easy to answer but if I'm looking at art someone obviously made with no credit, that alone is a big red flag for me. I'll also probably be curious or mad enough to try to use google to locate the original (you can take a pic with your phone if using a physical book to do this), track my way to the actual artist and contact them to ask them directly. Often stock images are used, in which case there's nothing to really research, as those usually come with a license and that's OK. Also, A.I. generated art is fucking everywhere now, too, and that's also a possibility. Either way, it is always sus to find a book with art or images of any kind with no credits, even a simple istock or shutterstock credit.
8. USE AN A.I. DETECTION PROGRAM
Personally, when it it comes to books and content on Witchcraft at least, I find the previous ways I've already listed to be fairly effective. I'm strictly talking about A.I. in the context of Witchcraft, not college papers or technical writing, so these tips will largely be sufficient for most everyone if you are looking to spot A.I. within this category of content.
With that said, if these are not sufficient or you want to step outside to other categories of writing, there are many programs you can find online that are exceptional at spotting AI-generated text.
One down side I've found on many of these sites is that often they are the same ones offering A.I. generated tools to create that text or content. I worry that every time they are used to detect A.I. text they are learning how to do it better so their own A.I. generated content gets harder and harder to distinguish between human and A.I.
If you want to go this route, I encourage you to do your own homework and try to find one that you feel good about using.
FINAL THOUGHTS
Not every red flag or tip here is a fool-proof way of deciding whether or not something has been written by A.I. There are real writers out here publishing their content that could be confused with A.I. generated content. Some writers are just not very good writers. Some writers are lazy and have poor ethics or research practices. Some writers are reclusive and like to stay out of the public eye. Some writers do a shit job at self-publishing and editing. This is where critical thinking skills are necessary, so use them! It may not always be clear, and the only thing we can do is try our best and that's all I'm asking.
I'm sure there are more ways to spot A.I. than I listed here. I would love to hear from you if you know any!
In Light & Dark,
Blog posts are available to access free for 30-days. To access older blog posts, please purchase a subscription plan, (plans start at just $1.99/mo) that fits your needs. Thank you for supporting my work!