My Day Job is in Information Technology, now increasingly focused on AI, of course. For my Side Job, though, I’m a professional author. I write Science Fiction and Fantasy stories, mostly at novelette or novella length. My work is regularly published in anthologies and through small presses like Tannhauser Press.
I don’t use AI to generate my stories. I enjoy writing, I’m good at it, and I’m always going to write my own prose. But, let’s be honest, there’s only so much Dave to go around, and only so much writing time left after the Day Job and just, you know, having a life.
I couldn’t help wondering if AI could assist writers like me in other ways. Writing, the physical act of setting down new words for a story, is only one aspect of this creative art we summarily lump under the umbrella of “writing.” Before that, there’s often a whole lot of development work that most people don’t think about (especially for SF and Fantasy stories).
Here are some typical developmental tasks:
- Ideation: Generating ideas for stories.
- Research: Doing research into topics related to a story, including things as diverse as spaceship design, alien planet design, historical research, crime solving techniques, history, theories of government, etc.
- Worldbuilding: Envisioning a whole new environment, such as a world and its history, as in Game of Thrones; or a realistic spacefaring culture, as in The Expanse; and more.
- Brainstorming: Developing details for a particular topic area, such as the design and characteristics of spaceships in The Expanse, include acceleration, weaponry, sensory capabilities, etc.
- Outlining: Assisting with “beat sheets” or outlines for stories.
Honestly, that development work can often consume more time than the actual writing itself.
The Setting
I had been doing development work for quite a while on a hard SF setting. Hard SF is a term that refers to an SF story with science that is as accurate as possible. It’s like The Expanse vs. the alternative, which would be something like Star Wars, which has an exceedingly tenuous relationship with any sort of science, especially physics. That doesn’t mean I don’t like Star Wars (well, with exception of that last trilogy), but I don’t really watch it for the science, anyway.
My SF setting is a generation ship, which is a spaceship designed to travel at slower-than-light speeds for many generations to reach a far-off stellar system while carrying a large number of passengers. In other words, a journey without cryogenics to conveniently let the passengers sleep through the centuries. Entire generations will live and die aboard the ship while it’s traveling between the stars.
Most of those passengers will live in a large air-filled cylinder 100 kilometers long, with a radius of ten kilometers. The cylinder, referred to as a habitat, is spinning on its long axis fast enough to simulate about 75% of Earth’s gravity. People live on inside surface of the cylinder. A spindle runs down the center of the cylinder; the central 80% of that cylinder is the Sun Tube, which provides both light and heat during the day. To simulate sunset, the Sun Tube gradually dims to about 10% of its output. At dawn, output similarly rises to the normal daytime level.
For background, the ship was badly damaged when war broke out while it was leaving the system. As a result of that damage, the habitat spinning came to a relatively violent stop, leaving the entire cylinder in zero-g. At the same time, mutiny crippled the ship’s crew. During the fighting, passengers were permanently locked out of all computer systems. Additionally, access from the habitat to crew areas was halted by expediently opening all connecting corridors to vacuum and “bricking” (rendering unusable by self-destructing the electronics) all space suits within the cylinder. Finally, the ship’s AI was given directives to prevent it from interfering with the crew conflict.
The habitat has now been isolated for over a thousand years. The ship has long since passed its original destination without slowing down. And the habitat has been in zero-g for this entire time.
That’s my setting. And that’s my context for brainstorming, too. What is the habitat, now known as Heaven, like after a thousand years? What’s the weather like? How have people survived? What cultures have arisen?
Let’s find out.
Brainstorming with Claude AI
To facilitate my brainstorming, I looked at Anthropic’s Claude, which features three state-of-the-art LLMs (Large Language Models):
- Claude Opus: A high-performance model specializing in complex tasks, advanced reasoning, and coding, often used in professional workflows.
- Claude Sonnet: A model that balances speed and intelligence, making it ideal for high-volume enterprise tasks.
- Claude Haiku: The fastest and most efficient model, designed for quick, lightweight tasks.
For my purposes, I thought Claude Sonnet was the best fit. The current version at the time of this exercise is Sonnet 4.6, which was released in February, 2026.
Most of the text in The Setting section above was part of my context for Claude. I also described my role as being that of an author, with the goal of fleshing out the setting for use in stories.
What ensued was a rather intense four-hour session of 31 consecutive prompts and the generation of around 55K words of content related to my setting. Brainstorming included topics like:
- Weather: What are the weather patterns and air flow within the habitat? Can you have rain in zero-g? What about lightning?
- Everyday Living: What is it like to live in the habitat? How do people dress for zero-g, or cook food in zero-g?
- Shelter: Where do people live? 1) Wall towns, which are towns of wooden construction anchored to the cylinder wall; 2) Wheel towns, which are small, free-floating towns built on the inside of a wooden ring and spun up by pedal-powered propellers; 3) Trading ships, and 4) Farms on floating islands of dirt and rock.
- Tech Level: The tech level available to the survivors is 1850s, but without gunpowder, major manufacturing, or widely available metals.
- Trade: How does trade work? Among other things, there are trading airships with routes that see them visiting many communities within the habitat.
- Economy: How does the economy work? Do they have any sort of currency? What about banking?
- Pirates: How do pirates fit into the picture? (If you have trading ships, you have to have pirates. It’s like a rule).
- The Law: How does law enforcement work?
- Farming: How does farming work? What animals have survived and/or thrived in zero-gravity? What are the cultural implications of the habitat’s farms?
- Religions: What religions exist within the habitat?
- Funeral Rites: Since the habitat is a closed system, how are dead bodies returned to ecosystem?
- Childbirth: Without gravity, how does childbirth work? Answer: Have your baby in a wheel town with simulated gravity.
- Metal: Do they have any metal devices? If so, where do they get the metal? Are there miners? Or people who salvage material from the wreckage of the old world?
- Entertainment: What do people do for entertainment? Music? Do they have traveling theatrical shows?
- Military Conflict: Do wars work? If so, how? What about pirate hunting? What weapons are available (and useful) in zero-g? What do people fight over?
- Sports: People have always been competitive, so what sports do the survivors have?
I know that’s a bit of a jumbled list. While it covers many of the topics that came up in the session, the list items don’t correspond one-to-one with the prompts. The real takeaway is that during this intense collaborative session, I worked with Claude and got answers to ALL those questions. Answers that I can use in the stories I plan to write in this setting.
But bear in mind, I’d already done a good bit of context development for the setting already. And those 31 prompts were essentially a conversation. I prompted, elicited details, chose which areas to refine further, honed in on areas of specific interest to me for stories, etc. It felt a lot like brainstorming with a personal assistant who has terrific research skills, but who sometimes forgets things or goes off on a little bit of tangent.
Conclusion
The AI wasn’t perfect. There were a few things that Claude got wrong, that I thought were impractical and thus unlikely in the setting. There were also a few cultural nuances that I was going after that Claude whiffed on, notably the culture of traders, which I felt was likely to have matriarchal tendencies given the world constraints.
Nevertheless, despite those minor issues, the session was impressive. I assure you, while there wasn’t anything developed in that AI conversation that I wasn’t capable of developing myself, it would have taken me much more time to do so. I’d estimate that by working with Claude, I probably accomplished a 100 hours of development work in just 4 hours. And bear in mind, I’m only a part-time writer, so those 100 hours would have been spread across weeks or even months (because I’m also juggling writing-related tasks such as writing other stories, event appearances, interviews, anthology invitations, etc.).
Future Steps
The content generated during the session was conversational. It’s not particularly well organized, so it’s not really publishable even as a freebie “Setting Bible” for myself or my fans.
An interesting challenge would be to develop the prompts necessary to transform the conversational content into a more organized and useful reference work, such as the aforementioned “Setting Bible.”
About The Author
