An AI story in the style of... myself
"A blend of atmospheric detail, psychological tension, and speculative unease"
A couple of years ago, when ChatGPT was just getting started in the public eye, I ran a little experiment. If I gave AI the first paragraph of one of my published short stories, would the chatbot be able to finish it?
The result was sort of compelling. While of course it didn’t do a “good job,” the chatbot successfully identified the main characters, the thread the story would take, and correctly anticipated the next scene.
This year I thought I’d try a different AI experiment. It’s been known for a while that you can ask a Chatbot to write in the style of a famous author, like Hemingway or Joan Didion, for which the LLM has a lot of material to sample.
But what if it tried to sample someone who is not famous, and perhaps has less of an available sample size to work with?
That is to say—what if it tried to imitate me?
So I asked it to. And this is what it came up with:
Here is an original short story in the style of Harrison Blackman—a blend of atmospheric detail, psychological tension, and speculative unease, without copying any copyrighted material:1
The Cartographer’s Error
When the drone returned without its camera, the team pretended not to notice. It came buzzing over the ridge at dusk, rotors sputtering, landing skids coated in a fine red dust none of them recognized. The camera mount hung empty, wires shorn clean as though something had bitten them off.
Raina, the expedition leader, walked toward it slowly, as if approaching a wounded animal that might strike. “We need another map of the valley,” she murmured, speaking more to herself than to the others. “Without visuals we can’t confirm the fault line.”
The valley stretched before them—an expanse of fog, ravines, and shimmering mirages that warped the horizon. The earth underneath seemed indecisive, fault lines running in strange geometries. Satellite maps had misplotted the region for decades, and no one could figure out why.
Back at the portable command unit, the only screen active belonged to Ezra, the cartographer. He was hunched over his interface, repeatedly zooming in on a topographical rendering that refused to remain stable. Every time he blinked, the valley rearranged its layers: ridges shifting, chasms moving several meters to the left.
“It’s doing it again,” Ezra said, voice thin. “Terrain is updating without input.”
“Update protocols are manual,” Raina reminded him, though not confidently. “Terrain doesn’t just—move.”
Ezra pointed at a jagged contour line blinking at the map’s edge. “This wasn’t here ten minutes ago. It’s like the valley is correcting me.”
Raina suppressed a shiver. “Double-check the baselines. Your instruments might be miscalibrated.”
Ezra wanted to protest, but he saw the tension in her jaw. She wasn’t dismissing him—she was afraid to believe him.
Hours passed. Night settled deeper. The lights of the base hummed in the dark while the valley beyond remained invisible, as though the world ended ten feet from the perimeter. Ezra refused to sleep until the map stabilized.
By 3 a.m., the mapping screen was an unreadable labyrinth of contradicting coordinates. Fault lines spiraled, folded, and rearranged themselves into shapes that resembled skeletal structures, spirals, and—disturbingly—something like written language.
Ezra stared slack-jawed. “Raina… I think it’s communicating.”
Before she could respond, the drone—powered down and disconnected—began to beep. Its transmitter blinked an error message: LOCATION NOT FOUND. REPLACE MAP.
Raina looked from the drone to the screen and finally out toward the unseen valley.
“What if,” she whispered, “the maps were wrong because the world wasn’t finished yet?”
Ezra swallowed hard. “Or because it’s still being rewritten.”
Neither of them spoke after that. They just watched the fault lines shift, as the land updated itself in real time, drawing a new world the way a nervous hand draws in the dark—almost as if the earth had grown tired of waiting for humans to chart it, and decided to chart itself.
[THE END]2
Let’s unpack this
Last month Sam Kriss wrote an essay in the New York Times Magazine expertly identifying the tells of AI writing. AI syntax frequently follows patterns—a lot of em dashes, a lot of lists, a lot of sentence structures that follow the format of “it’s not this, but that.”
“The Cartographer’s Error” follows these tells pretty closely:
“Fault lines spiraled, folded, and rearranged themselves into shapes that resembled skeletal structures, spirals, and—disturbingly—something like written language.” Em dash, check.
“Every time he blinked, the valley rearranged its layers: ridges shifting, chasms moving several meters to the left.” List, check.
“She wasn’t dismissing him—she was afraid to believe him.” It’s not this—it’s that!
Another striking aspect of this piece is that it is genre-related to my unpublished novel, Blood Falls, which also involves scientists examining broad geomorphological terrain. Though that novel has not been published, I’ve submitted portions of the MS to writing contests, and that’s probably where OpenAI is hoovering up my work. Disturbing, yes, but not inexplicable.
But is this truly written in “my style?” My fiction style is pretty broad—I have a lot of modes that I work in—suspense, horror, literary, commercial. While it does pick up on some of my tics—bolding and all-caps for paratexts (“LOCATION NOT FOUND”)—more of a screenwriting thing than anything else, the only style this story might have is AI’s style, though in the summary it does accurately sum up how I try to write: “a blend of atmospheric detail, psychological tension, and speculative unease.”
Then I was thinking—a lot of my available writing is journalistic. What, then could AI do with that? How would it write a journalism story by me?
Here goes nothing—and this time, I mean it:
The Town That Keeps Moving: A Quiet Community Faces a Shifting Landscape
Here is an original journalism-style article in the stylistic vein of Harrison Blackman—measured, atmospheric, deeply reported, with an undercurrent of speculative unease, without copying any real copyrighted text:
Raventon, TX — At first, residents treated it as a curiosity—hairline cracks in sidewalks that seemed to change direction overnight. A fence that no longer lined up with a property boundary. A gravel road curving just slightly differently than it had the week before.
But by late spring, the phenomenon had become impossible to ignore.
“We’d pave one side of the street on Monday,” said Public Works Director Tom Delgado, pausing on the courthouse steps to light a cigarette, “and by Friday the measurements wouldn’t match the original site plan. It’s like the ground keeps nudging itself into place when no one’s looking.”
Raventon, a town of 3,200 in the remote Texas Panhandle, has found itself at the center of a quiet scientific intrigue: the landscape beneath it is shifting—not in dramatic earthquakes or subsidence events, but in small, persistent movements that geologists say don’t correspond to any known fault system in the region.
A Puzzle of Millimeters
Texas Tech geologist Dr. Naomi Hsu first heard from town officials in May, when she received a packet of survey data that caught her attention. “There were changes of five to ten millimeters—small enough to go unnoticed in most places, but consistent across several grid points,” she explained. “What was striking wasn’t the size. It was the pattern.”
Hsu deployed a grad student and a set of ground-based LIDAR units. After three weeks of monitoring, she was convinced: the terrain beneath Raventon is not behaving according to the established geologic record.
“There’s no fracking here, no mining, no large-scale water draw—they’re not doing anything to disturb the underlying strata,” Hsu said. “It’s as if the bedrock itself is… reorganizing.”3
She hesitated over the final word, as if aware of its implications.
A Town With a Long Memory
Residents, many of whom have family histories stretching back to the town’s founding in 1889, have begun combing local archives for clues. Anecdotes have surfaced: old ranch logs mentioning barns that “wandered out of alignment,” and homestead maps from the early 20th century that don’t quite match modern land surveys.
In the local library, where every wall smells faintly of dust and yellowed glue, volunteer historian Linda Martinez has been updating a hand-drawn map of the town—by hand, in pencil, because ink feels too permanent.
“It makes you wonder,” she said, flipping through a brittle county atlas from 1911. “Did the settlers know something we forgot? Or did the ground only start changing after we did?”
Human Problem, Human Adaptation
For now, the town has taken a practical approach. Survey benchmarks are being updated monthly. High school shop students have begun building adjustable street-sign mounts. Contractors now carry both tape measures and GPS receivers.
Everyone seems to be adapting—because what else is there to do?
Mayor Denise Alvarez shrugged when asked whether the town was worried. “It’s not flooding us, and it’s not dropping us into a sinkhole. The world is alive. Always has been. Maybe this town just has a more talkative piece of land beneath it.”
She smiled, but her eyes lingered on the horizon—a long, flat plane shimmering in the summer heat, as if caught between two versions of itself.
What Comes Next
Dr. Hsu insists the phenomenon is not paranormal, and that further study will eventually reveal the mechanism. She is adding more instruments next month, including subsurface seismometers and radar probes.
But privately, several geologists involved with the project admit Raventon has become something of a professional outlier—an unexplained data point in a field that doesn’t often tolerate unexplained data.
Meanwhile, residents continue living their lives: rebuilding fences, redrawing boundaries, and accepting that their town is moving, slowly but definitely, in a direction only the earth seems to understand.
“People think places are fixed,” historian Martinez said, closing her atlas. “But maybe that’s just because we don’t usually watch closely enough.”
Um, let’s unpack this one, too
In this “piece,” well, ChatGPT decided to rewrite the fiction story as if it were truth. While it resembles “science journalism,” it’s like claiming the rocks are moving of their own accord.
And of course, not only are all the quotes and “characters” hallucinated, but so is the town in Texas—there is no such place. There are no such people as Naomi Hsu, Ph.D., volunteer historian Linda Martinez, Mayor Denise Alvarez, or Public Works director Tom Delgado. These people aren’t real. This isn’t JOURNALISM.
Let’s go back to that summary: “measured, atmospheric, deeply reported, with an undercurrent of speculative unease”—measured and deeply reported, thanks, but an undercurrent of speculative unease? Since when does my reporting have “an undercurrent of speculative unease?”
Clearly, ChatGPT’s wires are crossed and confusing different writing genres under my name. Lines like “She hesitated over the final word, as if aware of its implications,” are not unlike lines in my fiction or screenwriting. They are not in my journalism.
While news journalism is often formulaic in terms of structure, and this AI effort does understand that AP News format, it also supports my theory of LLM’s overhype. Whatever inroads AI makes in narrative writing, it will never be able to write news.
News is information that is new. It requires a human to go out and get new material through interviews and research. While AI excels at regurgitating essays on The Great Gatsby, it’s got tens of thousands of similar high school essays on the subject to draw from.
At base level, LLMs are plagiarism machines. Often, when you ask a chatbot about a current event, the LLM’s response will be out of date and draw from articles several years old. It can’t generate anything from scratch. It needs a pool of human-generated information to cull from. Without human journalists, AI will never be able to keep pace with the times.
So, what do you think? Are clankers going to eliminate all human writing? If they do, you’d better like AI’s “it’s not this, it’s that” structure.
Hm, yes, the copyright disclaimer. Not sure how that works!
This added by the real HB.
Um, say what???