Writing and AI

I think a lot of people are very afraid that AI is going to take creative jobs away, and we won’t be able to tell the difference between human content, and computer-generated content.

People are already using AI to spit out novels and create beautiful works of art.

I think there’ll be a perspective shift, soon, where things made by human hands are valued for the effort, the life, that was spent creating them. Machines are efficient. Humans—artistic. In the same way we appreciate beautiful calligraphy, or paintings, or anything handmade, we will always value effort.

Artificial Intelligence will create billions of stories and images, flooding the world, but it will make the handcrafted work of art all the more special.

I have had some amount of fear thinking about computers writing books that are “better” than mine. LLM prose is top-notch, it can shape and sculpt when wielded well. But there will always be something special about the human work of art.

A machine can create a blade, but there is much more value in one hammered and shaped by human hands.

You can 3D print a statue, but it won’t compare to the glory of a real stone sculpture, carved by hand.

There’s also a continuum here. I said, “by hand.” But to chisel a statue from rock, you need… a chisel. A hammer. Many more tools, I’d bet.

In the same way I write with an amazing computer, keyboard, and a “word processor”.

We had similar reactions to those things—people upset at those who use typewriters rather than those who write by hand—with a pencil. Those who told stories from memory, with their mouths only—those are the real OGs, I suppose.

But my em dashes are my own. I will never use AI to write for me. I love my words. If I make a typo I’ll correct it, and I do make the occasional grammar error. Writing my first novel I learned the difference between “wretched” and “retched”. The first is wretch-ed, like a dirty homeless person with a humped back. The second is vomiting.

And many more things.

I use AI quite often, I actually work for Data Annotation, a company that gathers training data for a bunch of AI models. It’s a lot of fun, and I get to read, write, and learn every day.

I use ChatGPT for all manner of questions, organization, brainstorming, venting, and general understanding. It’s amazing. But I recognize that I can’t let the machine think for me all the time.

When it came time for me to edit The Neighbors, I thought I’d do a great job. I have a good eye for catching mistakes, and I have, obviously, a computer at my disposal. But it was much harder than I expected. I got the story straight. I got my chapters straight. I got my scenes straight. I thought my prose was good, and I was feeling good, feeling polished. So I sent the story to my beta readers—Stan, Chris, Dirk, Trent, and Amanda. Stan sent me back the manuscript marked up to hell, finding a ton of things I was totally unaware of. Chris loved it, but pointed out that one character seemed to come out of nowhere because I had renamed them ¾ of the way through the story and forgot to fix the first three quarters.

That’s my father and mother, respectively.

So I dove back in, fixed what I could, did a sentence level edit all by myself, trying to perfect my prose with the Chicago Manual of Style and Strunk and White right next to me. That edit was a doozy, and after, I felt good. I felt like my shit was really smooth then. But, having learned my lesson, I thought I would reach out to some editors to see if anyone would take on the job. Turns out, finding an editor was easy, but the prices did not make me happy. I sent a snippet of the story, and all the editors came back with immediate fixes, as examples of how they could help me refine my prose—copyedit, proofread. That made me distraught! After all my work, however many edits, my own copyediting and proofreading, and I was still making mistakes. And fixing them was going to cost me.

But, I have ChatGPT, I thought.

ChatGPT could copyedit. I could stick in a scene, and ask for COPYEDITING ONLY. DO NOT REWRITE, ONLY OFFER SUGGESTIONS. FOCUS ON SYNTAX, GRAMMAR, CHARACTER VOICE…

I asked ChatGPT to make me a prompt that would accomplish that goal, and here’s what it spat out:

Role: You are a professional copyeditor performing a final proofread.

Reference: Follow The Chicago Manual of Style, 17th edition.

Scope & Priorities (in order):

·         Redundancy or repetition of words/phrases

·         Consistency of narrative & character voice

·         Glaring grammar, punctuation, or spelling errors

·         Unclear or ambiguous sentences/paragraphs

·         Strictly do NOT: rewrite or paraphrase entire sentences, alter style, change tone, or add material. Provide only pinpointed observations and corrections.

·         Response format: For each issue, use one of the five headings below and repeat as needed.

Typographic & Style Consistency

• Location – Quote just enough text or give ¶/line number

• Issue – What’s wrong and why (very brief)

• Suggested fix – Your Chicago-compliant correction ### Grammar & Syntax

• Location …

• Issue …

• Suggested fix …

 Dialogue & Voice

• …

Clarity & Cohesion

• …

Overall Notes

• High-level patterns or reminders (optional, keep concise)

General rules:

 • Keep bullet points short and actionable.

• If something is subjective, flag it as Consider: “Consider tightening …”

• Preserve the author’s wording whenever possible; suggest, don’t rewrite.

• Do not comment on sections with no issues.

• End your response after Overall Notes.

The “do not rewrite” was added because in my first couple tries, the model (ChatGPT) would try to spit edited text back out at me, which was NOT what I wanted. I wanted a copyeditor. A proofreader. I wanted ChatGPT to do exactly what a copyeditor would do—don’t write for me, but instead point out my mistakes, offer guidance on how to fix those mistakes.

I feel no guilt, no feeling that I didn’t put in the work or that I’m a fraud.

But I do I feel a bit of fear that people will consider it bad, and I can understand some of the reasons why. However for this first project, that’s the path I chose, and I’m happy with it. That’s probably why I’m writing this too, to come clean about the tools I used to edit.

Even writing this, the blue lines of Word’s editing software suggest I add commas and sharpen my language, but in these blogs I’ll write freely, leaving you with my natural language, rather than the polished version I’ll attempt to give in my books.

Thanks for reading, from my head to yours.