Who Wants to Be a Writer Anyway?

Plus: I tried ChatGPT and Now I Feel Dirty

I.

Who wants to be a writer anyway.
Spend your days tossing words
like skee balls at the arcade.
You miss the damn hole most of the time.
Maybe when you do spin it just right,
make the bullseye,
and the machine spits out your prize,
a precious string of coupons,
you can trade your words
for a plastic top
and an egg of silly putty.

Who wants to be a writer anyway.
Choose the wrong pen,
notebook, keyboard, font size
multiplatform writing suite
and fail before you start.
Butt in chair or it doesn't count
on your "real author" bio.
Open a vein, bleed on the page.
Is it supposed to hurt
like a knotty cramp
at that time of the month?

Who wants to be a writer anyway.
Start in in media res.
Write what you know, except adverbs
and prologues that tell, don't show.
Say ten hail marys
for the sin of head hopping.
The passive voice
should be avoided
like all bad writing advice —
or is it too cliche to say
kill your darlings?

Who wants to be a writer
when you can tweet instead,
riding the trends
on a surfboard algorithm.
Dancing on tiktokstagram
hashtag amwriting
while the four Tropesmen of the Bookopalypse
tell you there's no such thing
as writer’s block
when they can use AI
to compute the next philosopher's stone.

II.

The truth is, it only hurts when you’re not writing.

Generally, I prefer cat-assisted writing.

III.

Right up front, I have to state that I am ethically opposed to AI tools, specifically in the way they have been trained using work from writers and artists who were not given the opportunity to give or withhold permission for the use of their work, let alone given any kind of compensation.

But I am an ethically imperfect person and sometimes I cross the lines that I have set for myself, because I’m human.

As a writer, I’m not much concerned that AI is going to take my job.¹ The kind of writing jobs that AI will take are jobs you really don’t want as a writer anyway — I know, because I have done them. They pay fractions of a cent per word and are mostly regurgitating information that is already online. It is writing that is intended to fill spaces in digital marking plans because experts in SEO and analytics have told business owners they have to have “content” on their website in order to help the search engines funnel traffic to your site. It is soul-sucking, writing garbage for practically no money — doing that kind of thing caused me to slide into a creative depression that I am still recovering from.

I am more concerned about the noise AI-generated content is already making, and the trouble it’s causing for publishers, but I’m not an industry commentator and there are lots of good articles from the last year that you can find if you want to investigate this issue more deeply.

What I want to write about is about how, despite my ethical and professional objections to AI, I gave into the temptation to try ChatGPT, “just to see,” and the results of my limited experiment.

Here’s what happened:

I am working on the first revision of my novel-in-progress, and was stuck on one scene for a week or more, trying to find the right flow from point A to point Z, where Z was a particular decision that had to be made or the whole plot would be disrupted. I had the logical part of the scene mapped, but it was lacking the third rail that really brings drama alive, and I just couldn’t figure it out. Every time I opened the doc, I ended up rewriting the opening paragraphs over and over, hoping I would ignite the scene with the right word or phrase or image, but nothing was working.

I don’t know what made me think ChatGPT could help — maybe it had come up in conversation at work or maybe I’d read a post that day. All I know is that one night last week, I sat down and stared at my scene, and I wondered, “How would AI would write this?”

I created my account and entered my prompt, which was as spare as I could make it: Write a 500 word scene where character-X has to report to character-Y about character-Z, and then character-Y decides characters X and Y have to [do the thing]. It’s a fantasy setting. Character-Y is a little drunk.

ChatGPT spit out a scene that was … actually not bad. The narrative technique was solid, maybe a little dry — there was no interiority at all, no poetry, though it added in a reasonable level of description and detail. Of course, it was nothing I could have cut and paste into my manuscript. The characters didn’t talk like my characters, the setting doesn’t match my setting at all — I’d only given it bare bones and it made up the rest, which was kind of a relief because it eliminated any temptation to copy and paste any of it into my MS.

What was clear from that experiment was that ChatGPT had the same problem justifying the necessary decision-making that I was having — something about it wasn’t making sense.

So I told ChatGPT, “That doesn’t make sense” — it agreed with me and made a suggestion for an alternative. From there, we went on to iterate the scene half a dozen times, exploring different motivations and cultural expectations that might lead Character-Y to make the necessary decision, trying to find a flow for the scene that felt naturally dynamic and not forced.

I’m not going to lie — that iterative process turned out to be immensely useful to me. It was an efficient way to pinpoint the problem and experiment with multiple ways to solve it. ChatGPT could rewrite that scene in seconds, letting me explore different ways the scene might unfold without having to write each version myself. It was an assisted thought experiment, and when I shut down ChatGPT at the end of my session, I knew how I could rewrite the scene I had been struggling with. And I did!

I didn’t ask ChatGPT to write my scene for me. I never cut and paste a word it produced into my manuscript. The words that will appear in my novel will be 100% my own creation. Really, ChatGPT served as more of a writing buddy for me, letting me throw ideas around and offering interesting feedback, not supplanting my authorship.

But I still feel guilty about it.

Those ethical concerns about how the tool was built are still there, after all. No matter how useful it could be to me in the future as a way to quickly develop and road test creative ideas,² I know it would weigh on me to keep taking advantage of other writers whose work has been stolen.

So that’s the end of my AI adventures. Why am I confessing all this? I guess because I thought this was an interesting personal anecdote about something that is really impacting my vocation. And maybe I am just putting it out there that if there was an ethically-built AI writing buddy, it could be a useful addition to my tool box.

I guess this post is part of a series now about the intersection of technology and creative work. Read I 3D Printed a Unicorn Horn and Now I Feel Cheap on my blog.

Did you enjoy this post?

1. I am a full time marcom writer at a university and an aspiring novelist.

2. I don’t want it to actually write for me. Goodness, I LOVE writing and it’s pretty much the only thing I’m good at. Why would I delegate the best part?