What is Neogen?

Neogen is an experiment that is testing the current and future boundaries of human and AI co-creation of creative fiction.

More specifically, it is an enthralling science fiction mystery centered around a group of teenagers who uncover a baffling puzzle that tests their reality, they explore themes of identity and friendship. - co-created, written, illustrated and narrated by a 49-year-old dyslexic suburban dad and multiple AI platforms.   

A new chapter will be published each week - 100% free. 

As a lifelong comic-book and sci-fi aficionado, I've always had a galaxy of stories orbiting in my mind. The catch? I'm dyslexic with a dash of ADHD, which means while my head's buzzing with creative plots, getting them onto paper is like trying to text during a rollercoaster ride.

I've made my mark in the tech world, always first in line for the latest gadgets. Think of me as a tech magpie, always after the shiniest new thing.

By 2022, my new tech crush was generative AI. It wasn't just a professional intrigue; I started imagining ways to use it to funnel the whirlwind of stories in my head into something tangible I could share with my four boys.

Fast forward to 2023, and that's where the Neogen project comes in. It's my playground for testing the storytelling prowess of generative AI, like a mad scientist but with less hair-raising and more key-tapping. It's about pushing the limits of this tech while having a moderate amount of fun.

There are a few different phases to my workflow. 

Writing the book:

This project is a true collaboration, more than just a simple command to ChatGPT to write a book. So far, I've crafted 16 out of 35 chapters, aiming for a substantial progress before beginning weekly updates.

Here's how the magic happens:

1. I've created every character – their looks, family backgrounds, and insecurities.
2. I penned a rough outline of the entire story, from start to finish.
3. I divided the story into several parts, then into chapters, and further into scenes.
4. For each scene, I detailed the key plot points.
5. I then use AI to complete the prose for each plot point. My journey began with ChatGPT v3, but upon the release of v4, I switched to it for a fresh start due to its enhanced writing style.
6. Since ChatGPT doesn’t retain past inputs, I kick off each session by feeding the AI a summary of the story so far. This summary was written by ChatGPT at the end of the last session. 
7. I provide a scene-specific prompt based on the plot points I've outlined.
8. ChatGPT then drafts the scene, which I slightly tweak and add to the main manuscript.
9. Each chapter takes around 20 minutes to complete this way.
10. After finishing a chapter, I input it into ChatGPT  for a concise summary, which goes into a separate document. This is necessary due to ChatGPT 's token limit; it can't process the entire book at once.
11. This process repeats – write, summarize, and continue.

It's a rinse and repeat cycle, blending AI efficiency with human creativity.

NOTE: I have a paid account for ChatGPT which equals $20 a month. 

Illustrations:

There aren't many illustrations in the book, but I wanted a visual reference of the characters for the readers. All of these images were created using Midjourney. 

Audiobook:

I love audiobooks so I wanted to add a audible narration ASAP. All of the narration of the book is done by a AI at ElevenLabs.  It's the best I've found so far for text to speech. I simply paste the chapter content into the AI and it produces the narration using the voice I select. 

NOTE: I have a paid account for ElevenLabs which equals $22 a month. 

Writing the book:

ChatGPT operates on contextual cues, much like playing an advanced version of madlibs. It selects words and arranges them based on the likelihood that the combination will resonate with you. However, it doesn't truly comprehend the content it generates. So, for a response to match your specific needs, the right context is key.

For a book project, ChatGPT requires two types of context:

1. Book-Level Context: Since ChatGPT doesn't retain memory from previous interactions, each session starts from scratch, unaware of earlier parts of the story. This poses a challenge in maintaining a consistent narrative tone and character dynamics throughout the evolving story. To tackle this, I begin each writing session by inputting a summary of the story to date, which ChatGPT itself generated. This gives it the necessary backdrop. Ideally, I'd provide the entire story, but currently, the token limit (the maximum size of text ChatGPT can process in a prompt) restricts this, even in the paid version.
2. Writing Context: ChatGPT tends to be brief and to-the-point. If asked to write a book chapter, it might produce something concise, with minimal dialogue and description. To circumvent this, I divide each chapter into scenes and add specific qualifiers to each prompt.

Here's an example of how I might prompt ChatGPT for a scene:

Write a scene for a young adult audience, with the following plot points. Do not continue past the last plot point I give you in each prompt. Reference the content of the summary I’ve already posted as needed. Add lots of character dialog. Add internal dialog and motivation throughout the scene. Use descriptive adjectives. Make it dramatic. Take your time teasing out the scene. Include the following plot points but stop after the last plot point. 

• Devyn makes her way to her first afternoon class after eating lunch with her track teammates 
• Lennon is waiting for her and wants to talk
• Devyn is worried that Jay might see them, and suggests they duck into the band room next door.
• They go into the room together. 

Audiobook:

I've been pleasantly surprised by the simplicity of transferring a chapter of my book into an AI narrator, which does a fairly good job of bringing the text to life. However, one hiccup I've encountered is with character names. Since some of my characters have international names, the AI tends to pronounce them differently on each occasion, leading to a bit of inconsistency.

Firstly, it's important to acknowledge that as this technology rapidly advances, so do the ethical and legal debates surrounding it. Generative AI is trained on the collective works of billions, and while no creative endeavor exists in isolation, this type of work is a clear composite of numerous others' contributions.

My aim here is for enjoyment, exploring new frontiers, and to create something special for my children.