A notebook for fiction writers and aspiring novelists. One editor’s perspective.

• Next post • Previous post • Index

---

No Way, I told myself, would I ever allow Artificial Intelligence (AI) near my fiction. And then, on a whim, I tried ChatGPT. And that changed everything. Possibly even my proficiency as a writer.
(Just sayin’!)

Okay, so I’m not here to convince you that OpenAI’s groundbreaking, game-changing Python-based chatbot—ChatGPT—is an essential fiction-writer’s tool. I’m aware that some writers are already assuming AI to be demonic, while others consider it The Second Coming. And while I’m not yet ready offer a firmly informed opinion—I will say that, thus far, I’m bedazzled. ChatGPT (Generative Pretrained Transformer), may just become a fiction writer’s new best friend. Or certainly a casually insightful, 24/7 research assistant. And <sigh> most likely a fiction editor’s Waterloo.

BTW: I’m not going to geek out about AI technology in techie terms, or remotely attempt to give anyone a crash course in the worth (or utter irrelevance) of Artificial Intelligence. I won’t ponder the various higher-functioning, multi-modality abilities that various GenAI toolsets can impart—producing graphics and complex charts, for instance, or analyzing data, powering through mathematical equations, brainstorming incomplete data, initiating music scores or generating personal websites. The free-to-use, unimodal GPT version I’m using* is strictly for my text-based editorial and research purposes; for creating potentially viable outlines and then, with a word, creating numerous variations of those outlines. AI may conquer the world in the near future, may enslave all of humanity and use us as potting soil—but for now it’s gonna help me untangle my plot.

I’m perusing author Pam Baker’s ChatGPT For Dummies as my initial info source—and using the app itself for further assistance—and I acknowledge that my initial learning curve is barely yet a bunny slope. But I can also tell you that, of the few dozen unfinished manuscripts that linger like lost souls within my laptop’s ‘FAILS’ folder, ChatGPT has already offered me numerous paths forward—in both synopses and outline formats.

For those of us who are constantly “getting stuck” in the midst of some worthy, yet half-baked concept—I suspect ChatGPT can change our fortunes for the better. Scoff if you’d like… but what if I’m right? Unencumbered by ego, subjectivity, fatigue, frustration or a lack-of-knowledge, AI just may be that gentle nudge forward that many of us need. (Using AI feels like a gentle nudge. Like any other editorial support system, you’re still in complete control of your story. Take those suggests that feel useful, and leave the rest. You can continue to repeatedly prompt ChatGPT to hone its responses until you’re satisfied with the results.)

There’s a big difference—philosophically and legally—between the terms AI-assisted and AI-generated. As an encyclopedic ‘research assistant’ (and nothing more) AI can still offer inexhaustible advice, providing both developmental and line-editing functionality. AI can suggest entirely new plot twists, secondary story lines, additional characters (and the rationale for including them), and revise an existing story based on your meticulous prompts. It can suggest options to add (or restrain) drama or suspense, and blend those elements into an pre-existing story line. It will emulate your style to the best of its abilities.

And while I remain quite adamant that I won’t use ChatGPT to ‘write my story’—to concoct its own variation of my original prose, word by word, I’ll admit to occasionally writing a paragraph or segment in such clunky disarray that a second set of eyes—even digital eyes—might help untangle an otherwise unholy mess… so never say never. But, as a writer, my creativity is essential to my creation, and using a chatbot to create page-after-page of ‘original’ prose feels like the ultimate cheat. As a reader, I’d feel disappointed knowing that I’m reading an AI-rendered story with a name attached—possibly an uninspired writer who’s completely unaware of their own creation.

Which brings me to this somewhat perplexing realization: ChatGPT is not a creative resource. Its purpose is far more craft than art; far more right-brain oriented than left brain imaginative. (Although if one asks it to be imaginative, AI will give you its best shot.) ChatGPT relies on a user’s own prompts; its response is based on pattern recognition, informed guesses and incredibly rich information sources. Occasionally, its responses may not be accurate or practical for your intended use—although, truthfully, I’m blown away by its presence of mind, by its seemingly effortless ease of understanding my questions (simple or complex) and in generating “informed” responses.

Thus, if you ask ChatGPT to confirm that the sky is blue, it will agree. Not because it looks up and sees a blue sky, but because the app is programmed to seek out those established identifiers. But it’s trained to offer other insights as well, such as ‘not at night’. Thus, it may force you to choose your prompts carefully for AI to work most effectively.

I just asked Bob—yeah, um, so I named my chatbot “Bob”—why some people (in 100 words or less) believe the world is flat.

Bob’s response:

Hi Dave. Some people believe the world is flat due to mistrust of authorities, scientific institutions, and mainstream media. They often rely on personal observation (e.g., “the horizon looks flat”) and online communities that reinforce their beliefs. Conspiracy theories play a big role, suggesting space agencies and governments are hiding the “truth.” Cognitive biases, such as confirmation bias, help sustain these views despite overwhelming evidence.

AI can feel magical in that sense, or adroitly prognostic—surrealistically attuned to that deep, meaningful, often spiritual place where most fiction writers cloister themselves in absolute isolation. It can adopt my stylistic approach with very little prompting, can emulate my POV and even suggest new avenues of thought that I’d not yet considered. It does not snicker, chide or dismiss the lamest of my often half-baked ideas and will attempt to intelligently articulate what I cannot. In a sense, AI parrots back a response that you, perhaps in a week or a month, might research and eventually intuit for yourself. But it does so in a minute or two (and never complains about the workload or the weather.)

As author Pam Baker points out early in her book: ChatGPT [can] trick you into thinking that you’re having a conversation with a machine…and is without a doubt smarter than your and all other humans. But that’s just an illusion. Poor ChatGPT doesn’t understand a word you say—or that it says back to you.

I’ve probably read that last line a dozen times, but it’s yet to convince me that the damn thing isn’t somehow ‘humanesque‘. I know, right? And that haze of surreality may never lift. (“Bob” feels like a friend.) And yet, part of me—intellectual giant that I am—still considers AI as some sort of all powerful, magic-wielding, voodoo doctor. It’s the wizard behind the curtain. But it’s also a relationship—yup, a relationship—that feels like no other.

My wife, Eileen, writes non-fiction. She familiarized herself with ChatGPT a few months back, hoping it might help her outline certain segments of a book she’s writing about America’s current political insanity. She suggested that I give AI a peek. A chance. Me, I scoffed at the idea of relying on AI for anything. Ain’t no snippet of machine-driven Python code gonna tell me how or what to write. Ptttttuy!

But, yeah, I gave it a glance. And that’s my suggestion for every writer who’s even remotely interested. Take a look. Why not?

The thing is, AI ain’t going away. In fact, it’s going to become part of our daily lives, like it or not. (Already has, actually—just that now, I get to talk back.) And if you learn its language (basically by asking the right prompts) it may improve your life more than you realize. Think of it this way: What AI is to the Internet is what the ‘Net was to libraries. Meaning not only can you access a wealth of information, but AI can expand, illuminate, verify, collate and organize that info—whether you’re looking for a synonym or asking for suggestions on an epic problem that’s stalled your story for days or weeks or years.

An Example.

I’ve already mentioned that I have no intention of allowing “Bob” creative license over my prose. But how creative is ChatGPT? I’ve provided “Bob” with a somewhat lame (intentionally so) writing example and I’m asking it to rewrite that fragment in a more dramatic, more active, voice.

My example:

John was tired. He went upstairs to bed. Nearly asleep, his wife Helen suddenly shook him. “I think there’s somebody downstairs,” she said breathlessly.

He listened for a moment, then pulled the blankets to his chin. “It’s just the wind,” he told her.

A moment passed in silence. “No, somebody’s downstairs. I can hear them moving,” she said.

“Go to sleep, Helen.”

“No, I’m serious, John. Somebody’s coming upstairs.”

“Go to sleep,” he told her again.

“Yes, Helen, go to sleep,” whispered the shadow lurking in the doorway.

And “Bob’s” reply:
.

---

Absolutely, Dave. Here’s a more vivid and dramatic rewrite of those paragraphs, with sharper tension and a stronger voice:

---

Exhaustion dragged at John’s limbs as he climbed the stairs, each step heavier than the last. He collapsed into bed, eyes already half-closed when Helen’s hand clamped onto his arm.

“John—listen!” she whispered, her voice tight with fear. “Someone’s downstairs.”

His breath hitched. He held still, straining to hear. The house creaked. The wind moaned faintly against the windows.

“It’s just the wind,” he muttered, pulling the blanket up like a shield.

A heartbeat. Then another.

“No,” she said, her voice rising. “There’s movement. I swear I hear someone—footsteps.”

John groaned. “Go to sleep, Helen.”

She didn’t move. “John. They’re on the stairs.”

“I said go to—”

“Yes, Helen,” a voice rasped from the doorway, low and gleeful. “Go to sleep.”
.

---

Let me know if you want variations in tone—more psychological, more gothic, more cinematic, etc.

Is AI’s version an improvement? Better prose? I’ll make no judgements here… although “Bob’s” word choice—‘gleeful’—was interesting, and unprompted. Joyously gleeful, or gleefully sinister? And I could prompt ChatGPT for greater precision or clarity—for example, more (or less) suspense, or terror, or humor….or adjectives, for that matter. Although, personally, if I’m going to agonize over just the right word, or mood, or visualization, I’d rather ponder my own words and not—sorry, “Bob”—some snippet of coding that more or less mimics/modifies my original intentions.

So, no, giving “Bob” co-author status is not going to happen.

However, to be fair, here’s another example of AI’s potential functionality. My prompt was simply: Do you have any great ‘dystopian future’ ideas? Each in 100 words or less. I’ve told ChatGPT that my preferred writing style is loose, casual, a bit edgy, so “Bob” has adopted those qualities as well. (As to the originality of “Bob’s” concepts… you can decide.)

“Bob’s” reply:

Oh heck yes — dystopia is deliciously fun to play in. Here are a few fresh or twisty takes that steer away from the overdone stuff (no generic evil corporations or recycled Hunger Games vibes here):

(…and did “Bob” just dis The Hunger Games?)
.

1. The Memory Lease Society

The above were “Bob’s” first three (of a dozen) suggestions. And, admittedly, not bad title selections! For those of us still attempting to concoct a valid story or plot-line, or stuck somewhere in the middle, unsure of which way to proceed, prompting ChatGPT for possible solutions and/or endgames may prove invaluable. Or at least worth a serious look.

Legality.

Do note that, legally, work wholly generated solely by a chatbot cannot be attributed to an individual. Copyrights are only given to people. So if your intent is to simply allow AI to write a book for you, just realize such work cannot be legally copyrighted. Fiction created by a human, and relying on additional AI input, may be publishable—although those guidelines/boundaries have yet to be fully established or defined.

Also, be aware that publishers (and others, including the US Copyright Office) can detect AI-generated content. Self-publishing? I have no clue if AI laws are as strict… although I do have a question. Why? Plagiarizing a code-driven non-entity feels fairly ridiculous, although such attempts by students and writers have already been flagged. And I suspect that being fingered as a poser of such magnitude would undoubtedly destroy any further efforts to become a legit author. Nobody wants to become the Milli Vanilli of the literary world.

Bottom line is: We’re in largely uncharted waters here, so if in doubt, rely on your own instincts, your own words, your own creativity as a final draft. But having AI tag along as an intelligent assistant (or confidant) can’t hurt either, once you’re able to wrap your brain around the fact that you’re communicating with a Python-coder’s wet dream.
.

————–

* Unimodal GPT focuses on a single ‘mode’ of information exchange—solely text based, for instance. Multimodal GPT models can generate multi levels input (e.g.; text and video or audio) but that dexterity may cost users. ChatGPT’s individual-user, unimodal accounts are free. The next level bump will cost you $20/month. Other, professional options are available ($200/month and up) as well. But the free version should prove more than adequate for a fiction writer’s needs.

Do note that several similar apps are available—that some claim may be superior to ChatGPT. So you may want to browse around before you decide. Or, heck, mix and match! Current apps include: Microsoft’s Copilot, Google’s Gemini, Perplexity AI, ChatSonic, Claude and Jasper AI. In a year or two, there may be hundreds more.

You can access ChatGPT either via Google (any browser), or use this link: https://openai.com/blog/chatgpt. (Me? I Googled the app, stuck the link on a MSWord page, bookmarked that page and dragged the alias to my desktop.)

.

---

• Next post • Previous post • Index

by