Updated thoughts on ChatGPT’s “literary ending” to my novel scene

UPDATE (meaning that yesterday’s “last thoughts” weren’t quite that, apparently 🙂 ) – It struck me this morning (the next day, after posting about ChatGPT’s edit of my novel scene yesterday) that my own ending was kind of upbeat and optimistic versus the ChatGPT rewrite that ended on a down note. I was going to add an update at the end of that post, but it got so long that I just decided to give space to it in a new post of its own. 

Compare my original scene’s ending with ChatGPT’s revised ending.

My ending

Just for today, Jarrett decided in that split second. Just for today, he would enjoy being around her with no guilt and no worries about the future. Today was enough. Because in the end, he thought philosophically, moving toward the first mark as he and Taryn had practiced, “today” was all any of us really had anyway.

ChatGPT’s ending

Whatever this was between them, it couldn’t last. She was too young, their lives too different. When the cameras stopped, so would this.

The thought sat like a stone in his boot, impossible to shake.

My thoughts for future revisions

Remember, my original instruction to ChatGPT was:

Review and edit the following scene. My usual fault is over writing, especially using too much introspection, too much description, too much explaining, too much purple prose, for lack of a better word. Don’t over edit this scene but make any edits necessary to help it feel more literary and less like a trashy pulp novel.

It occurs to me that ChatGPTs more downbeat ending is doing exactly what I asked for: “to help it feel more literary.” Right? I don’t know why, but many of the “literary” short stories and novels I read are like that. Maybe it’s better artistically to have that “down” vibe going? But I have to think about 1) whether I want that vibe in my novel, 2) whether it fits Jarrett’s character and character arc, and 3) whether the dramatic arc of the scene, chapter, and storyline needs an “up” beat or a “down” beat in order for me to shape what I hope the reader’s experience of the novel will be.

Each of these scene endings would, of course, take place within the larger context of where the scene ultimately falls inside a chapter and inside the overall story. The scene itself is *sort of* modular, but not really. Because of the scattered way I work on this project (stealing time whenever I can from a very packed schedule), I’m writing many of my scenes as stand-alone units, but what this particular ChatGPT session has illustrated to me very clearly is that I can’t continue with the modular thinking for too much longer.

Why? Because a novel is an organic whole. It does not consist of interchangeable units. One scene flows into another, and if one scene ends on a down note, then somehow the next plot incident should either amplify that down note or reverse it, depending on the “journey” I want readers to take. 

When I begin to construct the larger story arc for this section of my novel, I need to think about the dynamics of how each scene interacts with the others. Not only should I be thinking about the reader’s experience of moving through the plot’s action, but I also need to think more about Jarrett’s character arc. Is he an optimistic, high-agency person capable of willing himself to think/be/act in a certain way?  Or, is he someone deficient in agency, too focused on negatives, with limited ability to envision much less move toward a better future? Where do I want him to start out as a character? Where do I want him to end up? And how will he move toward (and through) this change via the story’s plot events?

I don’t think I have anything super profound to add to this, except to say that it’s even more obvious to me now what the limits of ChatGPT are for writing. It’s the writer who needs to make decisions like the ones I described above. Once the writer has assessed the ChatGPT output and made revision decisions, they then need to communicate the new plan to ChatGPT and judge how well ChatGPT has executed the task before “recalculating” (the way GPS does when encountering an obstacle in the previously defined pathway) their “vision” for the overall piece again and moving into the next phase of crafting the text.

The writer remains at the center of the writing process, at least if the final text is going to be any good. ChatGPT truly is “just” a tool. A remarkable one, and quite fun to work with, but definitely not one capable of taking on all the work that a tired writer with a deadline might like to offload onto it.

Unknown's avatar

About Katherine Wikoff

I am a college professor (PhD in English, concentration rhetoric) at Milwaukee School of Engineering, where I teach film and media studies, political science, digital society, digital storytelling, writing for digital media, and communication. While fragments of my teaching and scholarship interests may quite naturally meander over to my blog, this space is intended to function as a creative outlet, not as part of my professional practice. Opinions are my own, etc.
This entry was posted in Creativity, generative AI, writing exercises, Writing, blogging and tagged , , , , , . Bookmark the permalink.

2 Responses to Updated thoughts on ChatGPT’s “literary ending” to my novel scene

  1. MELewis's avatar MELewis says:

    Once again, a highly instructive piece. I’ve been struggling with whether or not to use Chat GPT as a tool, but you’ve certainly demonstrated its value in at least advancing the draft. Fully agree that the writer’s work in terms of where the overall story is going and what ultimately works to move the arc forward cannot be replaced by AI. But it can be a great help! Thanks for sharing your process in such detail.

    Liked by 1 person

Leave a reply to MELewis Cancel reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.