AI vs The Creative World

Know the facts before you accuse someone… of ANYTHING.

This post has been shimmering in my head for a while… What is a while?

Somewhere around last week, I became the center of a THREADS hate fest from someone in the horror community because of a post I made in a horror review group. *The post has been taken down since then.

The said post asked reviewers “if you suspected AI being used by an author you know, would you mention it to them or DNF the book and move on.”

While many reviewers had said to DNF and move on, some “authors” took offense to it for more reasons than one.

1. I was accused of defamation of the author because I posted a picture of a section of text I suspected was written by AI.

2. They also accused me of being on a witch hunt.

3. The THREADS user stated that I blocked her on Facebook.

4. And FINALLY – I was accused of inputting an authors work into AI, therefore I was both stealing and infringing on copyrights.

Meanwhile, a user took her anger off Facebook, to THREADS and proceeded to share my name and post, so I could be attacked by an angry—but ignorant—mob. Therefore, I needed a moment to pen a rebuttal without being mad.

I want defend all four of those points, though the 4th one will be the longest.

1. First off, I never accused someone of AI – My question was if I “suspected,” plus I never mentioned the author or the book. On top of that, this author is not part of the horror community. Therefore, I never accused plus the reading demographic was completely different.

2. How is one book I was excited to read considered witch hunting AI writers?

3. I deleted the post when the comments got heated, so she couldn’t find it because it’s gone, not because her profile was blocked. I mean, search me to see. My profile is completely public because I am part of a publishing company.

4. NOW THIS IS WHERE WE ALL NEED TO STEP BACK AND LEARN ABOUT AI BEFORE YOU SPOUT LIES

AI – HOW DOES IT WORK?

The big giant world of AI gains its knowledge by corporations feeding information or “training” into an API (application programming interface), as well as information based on websites and public datasets. This means places like Facebook, Substack (yes, Substack), eBay, Amazon, Google, etc… allow AI APIs to collect data.

Private companies may enter data to enhance services to their employees, etc. Example: Hubspot helps sales people read data to help generate more sales.

BUT… A normal, everyday person, does not have access to training APIs for AI.

Front end userfaces like ChatGPT are not open platforms for inputting training data, except for session-based information. This information is stored during a session, then it is wiped from memory when the session is done. This means, I can have a conversation with AI, asking for help to write a paper or book in this one “session.” But tomorrow, ChatGPT will not have any recollection of yesterday’s conversation if I start over in a new session. I can not train AI about my book, that may never been finished, called “100 Shades of Bad AI.” So, it will not give Joe information about Cyan’s new book that is not written, nor published.

While companies like Meta have illegally uploaded copyrighted books — they are not being uploaded on the front end of ChatGPT by some malicious person with nothing better to do with their life.

Readers/Authors are eager and quick to accuse people of “training AI,” but yet they are stupid enough to believe that anyone can train the big massive computer in the sky. Think about it… if we (you and me and everyone reading this) could just load data into the FREE AI OUTPUTER, then we could destroy it just as easily. Everyone just take a word document, write gibberish, and upload. We’ll kill it before it gets started. COME ON… That is not how it works. And if you are stupid enough to believe it, then go ahead and trash me on THREADS.

ChatGPT does allow users to input a photo or text to generate information based off of your criteria. IE. you upload a photo of yourself, ask it to create an image of you in an action figure blister pack with XYZ. It will give you that based on data trained. AI isn’t always correct as I had done that with a clear photo of a 5-member boy band and I got 6-dolls in my blister pack. AI doesn’t always get it right.

MOVING ON…

In the Facebook post, when I was asked how I could tell a book was AI written, I explained to the angry crowd that it sounded flowery without substance, and the descriptions of the scenes lacked real humanized senses. If you look at “generic ChatGPT” text, just vomited out of the system, it reads that way.

INPUT: Describe a sunset leua in Hawaii on the beach.

OUTPUT: As the sun begins its descent toward the horizon, the sky transforms into a canvas of fiery oranges, soft pinks, and deep purples. The golden light spills across the gentle waves of the Pacific, turning the ocean into molten glass. Tiki torches flicker to life along the shoreline, casting dancing shadows on the sand. Musicians strum slack-key guitars and ukuleles, their melodies mingling with the rhythmic crash of the waves.

If an author put that into their book, I can tell it’s 100% AI-written, and I bet you see it too. Long sentences. Things that don’t go together like to “mingle.”

When my reason wasn’t good enough, I made the mistake of mentioning that I scanned 3 paragraphs into an AI-Detector and it came back with a high percentage. Again, I was accused of data training and loading the author’s work into the big world of AI for the world to steal. Again, no names-no titles—just 3 paragraphs.

Not every AI-written piece dropped in a AI-Detector will show as 100%, sometimes it comes out as 50% or even 0%. It’s is not perfect, at all. But if I read the Hawaiian beach luau in someone’s book, and I drop that in a checker and it comes back, I am going to ASSUME it is written with AI. Maybe not the entire book, I don’t have the time and effort to upload an entire book and check.

Inputting text into a AI-Detector does NOT train AI, nor does it save the text anywhere in the AI system. If inputting data to check for AI or to grammar check, then ChatGPT would know the answer to the next question.

The answer to that question is right here… LAST STAND AT DISCOLAND. And I will be the first one to say that I used ChatGPT to grammar check a chapter of Discoland before I posted it, just so I didn’t spell something wrong or put a comma in the wrong spot. This was for Substack and not for publication. I have an editor who likes to tear the work apart before it goes in a book or anthology.

Right now, we live in a mob-mentality world where people are so willing to go after someone—mostly because they think they know the facts because they read it on the web… probably an AI-generated summary which was incorrect. I don’t know if it stems from those people using AI to generate work therefore they are so easily offended by those who can sniff out the shit. I truly don’t know.

There’s just so much anger in the world right now. I think people need to take a step back – think logically because we have a brain that works. We don’t need a computer to tell us the writing doesn’t belong to a human person—or that author, because I know her writing… I can sense it/you can sense it – the computer just confirms it.

NOTE: No part of this post has been edited by ChatGPT or any other computerized editing platform for grammar errors or content. This is 100% horrible Cyan LeBlanc.