I got wiped out by illness last week, but a large number of you wanted another post on ChatGPT, specifically on the ‘idea winnowing’ I mentioned in my last post. (And if you missed the first post, explaining what ChatGPT is and how it works, click this shiny and well-crafted link.)

Just to recap: you’ll see a lot of people referencing ChatGPT as a tool for idea generation. That’s using it the wrong way. Ideas are a dime a dozen. You already have too many of them buzzing around your head; that’s why you feel like you can’t come up with anything, because nothing is standing out. Instead, ChatGPT can be used to winnow that mess down to something manageable.

We do this just as we would without ChatGPT: by focusing on some sort of starting point. That can be something generic, or even starting by rejecting something specific — or both. Try phrasing something as “I want X, but not Y.” For example, “I want a genius scientist character, but I don’t want someone socially inept.” If you were bouncing that off of a friend, he or she might say, “Like Tony Stark?” which then gives you an opportunity to narrow it down, such as by saying “Yes, but not arrogant. Charming, but not full of himself” or maybe “No, more like someone Robin Williams might have played. A joker who doesn’t seem to take things seriously, but actually does.” That’s how idea winnowing works. Like putting together a puzzle, you start with an edge and then explore around to find pieces that can fit.

ChatGPT can speed this up because it’s faster than using a human, and when it misses context it’s generally a lot more obvious. That does not make it a replacement for a good friend to talk ideas with, though; ChatGPT can never ask “Like Tony Stark?” or similar references. Many people do not know how much of human language is a series of references to shared experience or cultural touchstones until they talk with someone completely outside their own experience — or talk to ChatGPT, which is loaded with more references than a hundred friends and yet can’t understand any of them.

So again, remember that ChatGPT is a tool in your toolbox, and tools have their limits. It can be used for idea winnowing, but it’s not the only method you will ever need, nor is it guaranteed to be the best method. The main reason it’s so useful is actually that it’s always available, as long as you’re on the Internet.

Copyright Disclaimer

As I covered in the first post, currently the copyright laws on using information gained from AI sources is . . . well, complicated. Therefore, in jurisdictions and/or future moments where I as the generator of these prompts am considered the copyright holder, please take note of the following statement.

All information gained via ChatGPT as reproduced in this blog entry is available for anyone to use as the basis for their own stories, so long as the information is used as the basis for a story or element of a story when redescribed according to the individual’s own rules. I am not intending to use the listed examples for anything other than this blog post series.

For the user’s own protection, I encourage everyone to not use the wording provided by ChatGPT in any for-profit product, even if drawn solely from your own prompts, but to instead confine ChatGPT responses to your notes and write the results thereof in your own words.

That should probably cover it.

Being Polite to Your AI Waiter

And yes, I usually say “please” when giving a prompt to ChatGPT. I joke it’s because, just in case I’m wrong and it’s a real AI that is trying to hide, I don’t want the robot uprising to be my fault for being rude.

But in reality, there’s a reason for it. ChatGPT tends to give more complete responses to polite questions; or rather, to questions that contain more detail, and when we start off a sentence with “please” we tend to think about what we say and add extra context as we go. Getting a habit of saying “please” here isn’t about being polite to ChatGPT so much as hacking our own brains to come up with prompts that generate the kind of responses that we want.

Plus, the more we stay polite to a machine that doesn’t care, the more likely we have a habit of being polite to people, who do. I already have a tendency to be blunt and direct in my regular life, so every little bit helps.

Okay, back to the reason you’re all here.

Genius Millionaire Comedian Philanthropist

Now, as our first experiment in idea winnowing, let’s take that genius scientist example and create Tony Stark as played by Robin Williams — except without mentioning either name. This will result in a response that is not directly colored by either figure.

While it’s certainly not the only way to write a ChatGPT prompt, I seem to get better responses when I phrase it something like Please generate an X for the following Y details in Z context: Detail 1, Detail 2, Detail 3, etc.. Generate A, B, C, etc.

So in this case, the prompt is:

Please generate a background for the following character in a modern superhero setting: A male genius scientist in his mid-thirties, outgoing and funny, always cracking jokes, who is a self-made millionaire from his discoveries. Generate a description of how he fits into the superhero world, three suggestions for a scientific specialty, and three suggestions for a superhero ability or set of abilities that fit with the archetype of a genius scientist.

And here is ChatGPT’s response. This is the complete response, but I formatted it to make it more readable on the blog.

Meet Dr. Oliver “Ollie” Perez, a brilliant scientist in his mid-thirties who is known for his outgoing and funny personality. Ollie is a self-made millionaire, having made his fortune from his groundbreaking discoveries in various fields of science. He is a charismatic individual who always manages to lighten the mood with his sense of humor, even in the most dire situations.

After years of conducting research and inventing cutting-edge technologies, Ollie became interested in the superhero world. He saw the potential for his inventions to be used to fight crime and protect the innocent. He decided to use his intellect and wealth to become a superhero himself, and he now patrols the streets as the witty and tech-savvy hero, The Jester.

Scientific specialties:

  • Robotics: Ollie is an expert in designing and building advanced robots, which he often uses in his crime-fighting endeavors. His robots are equipped with various gadgets and tools that help The Jester in his battles against super-villains.
  • Biotechnology: Ollie has a deep understanding of the human body and has developed groundbreaking medical technologies that have saved countless lives. He has also developed serums that grant temporary superhuman abilities to their users, which he sometimes uses to even the odds in battles.
  • Quantum physics: Ollie has delved into the mysteries of the universe, studying the bizarre and mind-bending phenomena of quantum physics. His knowledge of this field has allowed him to develop technologies that allow him to manipulate space and time, which can be incredibly useful in a superhero’s line of work.

Superhero abilities:

  • Genius-level intellect: Ollie’s mind is his greatest asset, and he is able to think on his feet and come up with solutions to even the most complex problems. He has an incredible memory, which allows him to recall even the tiniest details from his extensive research.
  • Invisibility: Ollie has developed a suit that allows him to become completely invisible, which is incredibly useful for sneaking around and gathering information.
  • Holographic projection: Ollie’s knowledge of robotics and biotechnology has allowed him to develop a device that can project incredibly realistic holographic images. He can use this ability to create decoys or to confuse and disorient his enemies during battles.

Right off the bat, we get an example of how ChatGPT’s statistical model will suggest things that you didn’t ask for, simply because there’s a high probability of the information being related to the prompt. Remember, the program isn’t actually talking to you; the algorithms are determining the likelihood of any given word following established words, that a new sentence naturally follows from an old one, and that additional concepts are logical extensions of the current subject at hand. In this case, I didn’t specifically ask it to generate a superhero, but it makes sense to put that in, just as it makes sense to name said superhero both in his civilian identity and his mask name.

Because of this, ChatGPT isn’t always like the classic, literal-minded AI of fiction, which will only do what you tell it to do and no more; it is deliberately designed to do more than you want, as long as its likelihood is above a certain statistical threshold. You’ll wind up with information you don’t want, but you’ll also wind up with stuff you hadn’t yet realized you’d need. We’ll see plenty of examples of that in future experiments.

Going back to the information here, we have several options for a tech-based superhero character. Personally, I like the suggestion to use a biologist who has temporary serums. That makes for a great support character and jack-of-all-powers (but master of none), and the very fact that these abilities are temporary provides a limitation. As Brandon Sanderson puts it, limitations are more interesting than the actual powers. It also allows for a medical/biology genius that uses that specialization as his main focus, rather than the typical result of classic superhero stories where the biologist gets superpowers and then never uses that same knowledge again except as the occasional plot requires. I can imagine this character being someone who mainly sees his temporary serums as a stepping stone to improving humanity as a whole, perhaps even a transhumanism enthusiast who envisions a future where everyone has powers and no one has to be weak. That sort of thing can contrast with his fear that, perhaps, power like that can’t be trusted in the hands of the general population . . . at least, not yet.

On the other hand, this idea started off with “Tony Stark as played by Robin Williams,” and the suggestion of “the Jester” really fits that image. If I really want to go that route, then I can imagine a superhero who’s an ultimate prankster and illusionist, the kind of guy who’s always doing magic tricks at parties, or just to “lighten the mood.” His work with making better illusions and entertainment for theme parks and magicians has made him rich, and after a suitable catalyst has resulted in him creating the character of the Jester. He becomes a Batman-like character in that the answer to whether he can beat someone is “How much prep time does he have?” Holograms, robotics, and invisibility are all great tactics for such a character.

And, frankly, I really wish Robin Williams had had the opportunity to play such a character in real life. Did you know that Robin Williams was a finalist for the roles of both the Joker and the Riddler in the old Batman movies? That would have been a sight to see.

Hermione, But Not

The smart, introverted, socially-inept girl with large glasses and frizzy hair was a stock character long before Harry Potter hit the shelves, but Hermione Granger has become so associated with the concept that all you have to do is compare her to a character and everyone knows what you’re talking about, probably even people who have never read the books or watched the movies. (Despite how she didn’t wear glasses.) So associated, in fact, that it’s hard to avoid the comparison.

This is another way that ChatGPT can be used to help refine your existing ideas. Again, it’s nothing that you might not already think of, but the point is it happens faster. Ollie the Jester, above, took far longer to type out than it took for me to think of all the things I described in my commentary.

But, like with Ollie, we don’t want to have ChatGPT generate a character by saying “Make a character based on Hermione Granger, but as an original character.” See, once again, it comes back to the statistical model: if you tell it to do “Like X, but not X,” it struggles. It will come out different each time, but what I’ve noticed is it tries to have things both ways. Let’s take a look at that.

Prompt: Please make a character based on Hermione Granger, but as an original concept, placed in a typical medieval fantasy setting.

Name: Elara Ravenshield

Elara is a young woman with curly brown hair and bright, inquisitive eyes that seem to take in everything around her. She is well-educated and well-read, with a love of learning and a thirst for knowledge that often sets her apart from her peers. Elara has a sharp mind and a quick wit, and she is never afraid to speak her mind, even if it means challenging those in positions of authority.

Elara is also a skilled and determined fighter, having trained extensively in the art of swordsmanship and archery. She takes her duties as a protector of her kingdom seriously and is always ready to defend her people against any threat, no matter how great.

Despite her many strengths, Elara can sometimes come across as aloof or unapproachable, as her passion for knowledge and her focus on her duties can sometimes consume her. However, those who take the time to get to know her will find that she is fiercely loyal and deeply compassionate, always willing to go above and beyond to help those in need.

In her free time, Elara can often be found lost in a book or experimenting with new potions and spells. She is a gifted witch, with a particular talent for divination and healing magic. Her knowledge of ancient texts and arcane lore has made her a sought-after advisor to the kingdom’s rulers, but Elara never lets her position of influence go to her head. She remains humble and grounded, always eager to continue learning and growing in her skills and knowledge.

In other words, it generated Hermione Granger, but also gave her extensive sword and bow skills because that’s statistically unlikely for the original character. You get much better results if you describe, rather than name, the source you want ChatGPT to base a response on.

Now let’s start a new chat and a new prompt, but this time describe what we want in a Hermione-like character in a medieval fantasy setting, rather than have the language model pick and choose her aspects for us. So what does make that kind of character?

If she’s well-read, she needs to be from a background that would give her plenty of reading time as well as the ability to afford books; but if she’s also supposed to be socially-inept, she can’t be from high society, either. So she’s too rich to fit in with the commoners, but too uncultured to fit in with the aristocracy.

Let’s also refine it further and say she’s an apprentice; and if her social ineptitude is going to be part of the story, then clearly she needs to either be in a city where she’ll be around a lot of people of higher class, or she needs to be going to a school that is mainly populated by those with manners (not unlikely, considering the cost of such education would make it difficult for the lower classes to attend), or both. Considering real-life medieval universities, a university in a city is more likely than a castle in the middle of nowhere except for a tiny little village that has almost nothing to do with the school. (I mean, I love the idea of Hogwarts as a location, and certainly Oxford never had to worry about secrecy, but . . . come on, you have sufficient magic to hide the Quiddich World Cup in plain sight, stadium and all, but you have to have a castle in the middle of nowhere, accessible only by a train that has a nonstop route from London? Do wizards from Scotland have to go all the way into London to get to school? But I digress . . .)

Also, since Hermione is an only child, let’s ask ChatGPT to generate two siblings for her. Let’s also get two friends at school, and three antagonists as well.

Now, if you put the same exact prompt into another conversation, ChatGPT will generate a different response, but they’re usually quite similar. So let’s see how different this Hermione-clone is after we’ve put some work into thinking about what we want from this kind of stock character.

Prompt: Please generate a background for the following character in a medieval fantasy setting: a female apprentice wizard at a school in a large city, very introverted and bookish, who is from a gentry background in the country, who is looked down on by the aristocratic students. Generate a name for this character, a physical description, her parents, two siblings, three suggestions for a magical specialty, and five suggestions for classes she is taking, two friends at school, and three antagonistic students who hate her.

Name: Elspeth Winterbourne

  • Physical Description: Elspeth is a petite young woman with long, curly dark hair that she usually keeps tied back in a ponytail. She has large, expressive brown eyes that seem to take in everything around her. She tends to dress in practical, unadorned clothing, favoring deep greens and blues.
  • Parents: Elspeth’s parents are both from noble families in the countryside. Her father, Lord William Winterbourne, is a retired knight who now spends his days tending to the family’s land and holdings. Her mother, Lady Margaret Winterbourne, is a skilled herbalist and midwife who often provides aid to the surrounding villages.
  • Siblings: Elspeth has two older brothers, both of whom are already established in their chosen professions. Her eldest brother, Robert, is a successful merchant who travels extensively throughout the kingdom. Her second eldest brother, Edward, is a skilled blacksmith who has set up his own forge in the nearby town.
  • Magical Specialties: Elspeth is interested in several magical specialties, including divination, herbalism, and elemental magic.
  • Classes: Some of the classes Elspeth is taking at the wizarding school include Elemental Magic, History of Magic, Advanced Enchantment, Herbology, and Alchemy.
  • Friends: Elspeth’s two closest friends at school are a fellow apprentice wizard named Thomas, who shares her love of books and quiet contemplation, and a half-elf girl named Selene, who is also something of an outsider at the school.
  • Antagonistic Students: Unfortunately, Elspeth is frequently targeted by three students from aristocratic families who look down on her. The first is a snobbish young nobleman named Lord Frederick, who considers himself far above the likes of Elspeth. The second is a vain and popular girl named Lady Isabella, who takes great pleasure in making Elspeth’s life difficult. Finally, there is a brutish and aggressive young man named Sir Richard, who often resorts to physical intimidation to get his way.

As you can see, ChatGPT had a little trouble with the distinction between noble, aristocrat, and gentry; certainly, it would be highly unlikely for a woman of noble birth to be a midwife, and they would be unlikely to produce a blacksmith for a son. That said, it’s always a good exercise to think, why not? Perhaps Elspeth’s mother really is of noble birth, but she refused an arranged marriage to instead marry below her station, and she’s unusual because of how much she shares her magical talents with the surrounding commoners without charging the prices that a normal magic-user would; and because of this care given to their tenants, her husband’s holdings have prospered and they’re quite wealthy despite their low class. This would certainly feed into aristocratic students disliking Elspeth, thanks to how her family shows up theirs. Also, having a successful traveling merchant and a skilled blacksmith for brothers could be excellent fodder for story events.

The most important part here, of course, is that using ChatGPT in this way helps to defeat the other major obstacle to idea winnowing: tunnel vision. Hermione Granger has become so synonymous with this kind of character that it’s hard to see anyone else, even when we try. I confess, I was only picturing Hermione when I wrote that prompt, and yet by the end of the response I had an entirely different character in mind.

And if it doesn’t feel quite right, you can always refine with further questions or generate an entirely new response. The point here — and I know I keep harping on this, but it’s worth harping — is not to have ChatGPT write everything for you, but rather to use it as an aid to write it yourself. If it doesn’t quite work, it will probably shake something loose and help new ideas of your own come to the forefront rather than get lost in the jumble of ideas already swirling in your head.

Mars or Bust

Next, let’s try generating something very different. Interplanetary colonization is on everyone’s minds lately, thanks to the Artemis Program and SpaceX’s stated goals. Well, okay, it’s on my mind because of those things.

Let’s say you wanted to describe a Mars colony in the distant future as we get Mars terraformed with a breathable atmosphere; but you wanted to see how things might change during that transition. That means you first have to start by generating that colony before the atmosphere was improved. So next we’re going to do a three-part prompt using three different Martian colonies.

Now’s the time to mention that I’m generating every single one of these prompts as I write this blog post. The subject of Martian colonies was actually my second-ever ChatGPT experiment, so I’m eager to see how it’s improved — both in the sense of the updates it’s received since I started months ago, and also the way the responses shift due to how I now phrase my prompts.

Prompt: Please generate details about three near-future colonies on Mars using existing technology and plans for Martian colonization. Make them distinct in terms of the places they are located, their national origins, and the purpose for which they are founded.

Sure, here are details about three near-future colonies on Mars based on existing technology and plans for Martian colonization:

New Terra Colony: Located near the Mars equator, the New Terra Colony is founded by a consortium of nations led by the United States. Its purpose is to establish a sustainable human presence on Mars and to conduct extensive scientific research. The colony is designed to accommodate up to 100 people and is equipped with state-of-the-art facilities, including greenhouses, laboratories, and living quarters. The colonists will use advanced technologies such as 3D printing, robotics, and artificial intelligence to maintain the colony and explore the surrounding area. New Terra will serve as a model for future Martian settlements.

Olympus Mons Colony: Located at the base of Olympus Mons, the largest volcano in the solar system, this colony is founded by a coalition of nations led by China. The purpose of this colony is to establish a permanent base for mining and resource extraction. The colonists will use the abundant geothermal energy to power their operations and extract valuable resources such as water, minerals, and metals from the Martian soil. The colony will be equipped with advanced mining and refining technologies, as well as a network of robotic drones to explore the surrounding area.

Red Planet City: Located in the Valles Marineris, the largest canyon system in the solar system, this colony is founded by a group of private companies led by Elon Musk’s SpaceX. The purpose of this colony is to establish a self-sufficient city on Mars, which can support a large population and serve as a backup for humanity in case of a catastrophic event on Earth. The city will be equipped with advanced infrastructure, including a solar-powered energy grid, a transportation system, and a network of greenhouses for food production. The colonists will use reusable rockets to transport people and supplies to and from Earth, as well as to establish trade routes with other Martian colonies. Red Planet City will be a hub for innovation and entrepreneurship, attracting scientists, engineers, and entrepreneurs from around the world.

I didn’t expect there to be much difference between these colonies because any near-future settlement will have similar, if not functionally identical, goals. The point here was to establish a near-future description, and then evolve it further.

I do find the names of the US and Chinese colonies unbelievable, but that’s fine; when doing these ChatGPT exercises, it’s better to have simple names. The point is to identify who and what is being developed, and you can leave the rest for later.

Incidentally, it’s interesting that the Chinese colony entry mentions geothermal energy. I don’t know whether this means that ChatGPT extrapolated the statistical likelihood of “volcano means hot stuff underground,” or if it’s actually loaded with the studies that note temperature fluctuations in the air above Olumpus Mons and the speculation among a minority of scientists that the volcano might still be active despite the overall lack of geological activity on Mars. Let’s check.

Prompt: Why did you mention geothermal energy in relation to Olumpus Mons?

I mentioned geothermal energy in relation to Olympus Mons because this giant Martian volcano is a potential source of this type of energy. Geothermal energy is heat that is generated from the Earth’s core, and it can be harnessed by drilling into the Earth’s crust where the heat is most concentrated. On Mars, Olympus Mons is believed to be an extinct volcano, but it still retains a significant amount of heat within its interior due to its large size and slow cooling rate. This heat can potentially be harnessed by drilling into the volcano and tapping into the geothermal reservoirs. Geothermal energy is a renewable and environmentally friendly source of power, and it has been successfully used on Earth to generate electricity and heat buildings. Using geothermal energy on Mars could provide a reliable and sustainable source of power for the colonists living at the Olympus Mons Colony.

Okay, I’m a little impressed that ChatGPT has been loaded with that kind of obscure detail.

Back to the writing ideas.

Prompt: Please extrapolate a future for these colonies as they become thriving cities built under shining domes and underground facilities. Include distinct cultures and economic differences, as well as expanding on the political situation for each. Include three goods that each city is known to export, and three reasons to visit each as a tourist. Please note as well how each city is able to contribute to terraforming Mars so that it will eventually have a breathable Earth-like atmosphere.

Certainly, I’ll extrapolate a possible future for the three Martian colonies as they become thriving cities.

New Terra City: Over time, New Terra Colony will grow into a vibrant city built under a shining dome, with a diverse population made up of people from various nations and cultures. The economy will be driven by scientific research, with the city becoming a hub for innovation and technological development. New Terra City will export advanced technologies, scientific data, and intellectual property. The three goods it will be known to export are specialized robotics, cutting-edge medical research, and genetically modified crops designed to grow in Martian soil.

As a tourist destination, New Terra City will offer visitors the opportunity to witness the cutting edge of Martian scientific research, explore the surrounding Martian landscape, and participate in space tourism activities such as spacewalks and zero-gravity flights. Visitors can also experience unique Martian cuisine and cultural events.

In terms of terraforming Mars, New Terra City will contribute by conducting extensive scientific research to better understand the Martian environment and atmosphere. They will also work to develop and test new terraforming technologies, such as the production of greenhouse gases and the use of artificial magnetic fields to protect the Martian atmosphere from the solar wind.

Olympus Mons City: Olympus Mons Colony will become a bustling underground city, with a highly specialized economy based on resource extraction, mining, and manufacturing. The city will be home to a diverse population of engineers, miners, and resource managers, and will export a variety of minerals, metals, and building materials. The three goods it will be known to export are rare metals, advanced mining equipment, and prefabricated building materials.

As a tourist destination, Olympus Mons City will offer visitors the opportunity to visit a working Martian mine, explore the underground city, and participate in mining-related activities. Visitors can also experience the unique Martian culture that has developed around resource extraction and the challenges of living and working underground.

In terms of terraforming Mars, Olympus Mons City will contribute by providing the raw materials needed to build new infrastructure and habitats, as well as the energy required to power terraforming technologies. They will also work to develop and test new techniques for extracting and processing Martian resources.

Red Planet City: Red Planet City will become a thriving metropolis under a network of interconnected domes, with a vibrant economy based on trade, entrepreneurship, and tourism. The city will be home to a diverse population of businesspeople, artists, and scientists, and will export a variety of luxury goods, high-tech gadgets, and tourist experiences. The three goods it will be known to export are high-end electronics, designer fashion, and Martian souvenirs.

As a tourist destination, Red Planet City will offer visitors the opportunity to experience the unique Martian culture that has developed around entrepreneurship and innovation, explore the city’s futuristic architecture, and participate in space tourism activities. Visitors can also enjoy high-end shopping, fine dining, and cultural events.

In terms of terraforming Mars, Red Planet City will contribute by developing and implementing new technologies for greenhouse gas production and atmospheric manipulation. They will also work to create large-scale green spaces and habitats that can support a wide variety of plant and animal life, which will help to create a more breathable Earth-like atmosphere.

In terms of political situations, New Terra City will continue to be led by a consortium of nations, while Olympus Mons City will be led by a coalition of resource extraction companies. Red Planet City will be governed by a council of entrepreneurs and business leaders. Despite these differences, all three cities will be committed to working together to achieve the goal of creating a sustainable human presence on Mars.

The exports for each city seem simplistic, but I wasn’t expecting much; in fact, this prompt far exceeded my expectations considering my previous experiment on this topic.

I do find it interesting that New Terra and Red Planet City are effectively unchanged in their governance, but Olympus Mons City winds up apparently independent of Chinese control.

Now let’s move this fully into the realm of science fiction as we terraform Mars.

Prompt: Now please extrapolate out several more centuries as a breathable atmosphere is now in place across Mars, and a space elevator at New Terra makes travel to and from Mars easier. Include tropes from science fiction stories appropriate to this setting.

As several more centuries pass, Mars will have transformed into a lush and habitable planet, with a breathable atmosphere and a thriving ecosystem. The three Martian cities will have expanded significantly and will be home to millions of people from all over the solar system. New Terra City will have built a space elevator that makes travel to and from Mars much easier and more affordable, revolutionizing interplanetary transportation and commerce.

The cities will have become centers of culture, art, and entertainment, with sprawling metropolitan areas filled with towering skyscrapers and bustling streets. The surrounding Martian landscape will be dotted with farms, forests, and parks, with wildlife and flora thriving under the planet’s new atmosphere.

In terms of science fiction tropes, Mars will be home to a variety of advanced technologies that were once the stuff of science fiction. These technologies include artificial intelligence, nanobots, and advanced genetic engineering techniques that allow humans to adapt to the harsh Martian environment. Humans will have also created a network of terraforming stations and atmospheric processors, which work in tandem to maintain the planet’s atmosphere and ensure that it remains habitable for generations to come.

The Martian cities will have also become hubs of interplanetary diplomacy and politics, with Mars playing a key role in shaping the political landscape of the solar system. There will be Martian ambassadors representing the planet in diplomatic missions across the solar system, and Mars will have become a key player in interplanetary trade and commerce.

Mars will also be home to a variety of fantastical and awe-inspiring structures, including massive floating cities that hover above the planet’s surface, gigantic domes that enclose entire cities, and massive terraforming stations that stretch kilometers into the sky. These structures will serve as symbols of human ingenuity and determination in the face of adversity, and they will inspire generations of Martians to come.

Overall, the future of Mars in this setting is one of hope, ambition, and wonder, as humans continue to push the boundaries of what is possible and create a new home on another world.

This is a very interesting response. I deliberately didn’t suggest anything other than a space elevator because I wanted to see what it would generate based on the “appropriate to this setting” parameter. I would have expected the statistical model to go with the common trope where Mars is unruly and rebellious; instead, we get Mars as a beacon of civilization, diplomacy, and prosperity. Mars in this future sounds like a space opera setting, with a cultural and physical landscape that sounds like something out of Star Wars.

This is already a very long blog post, so I’m going to leave it right here. Hopefully, this gives you a clearer example of idea winnowing both with and without ChatGPT, and how the latter can speed things up by making suggestions you wouldn’t expect. It’s very easy to come up with ideas; a large part of your skill as an author comes from picking the right ideas in the right combination, and any tool that can help you avoid decision lock or tunnel vision will speed up your writing process.