Experiments with ChatGPT

Why I won't be using an AI program to write fiction for me

Created on Midjourney

A brief introduction to ChatGPT

This post is going to be a little different. Feel free to ignore it, read part, or read all. Some of you may already be familiar with the concept of chatbots. They’ve become quite sophisticated, though they still have many flaws.

People are using them now to do everything from writing their school essays (not kidding!) to creating code for website widgets, to writing e-books for publication to creating courses to sell. One news outlet even used it to write news articles. That one blew up in their faces when the articles proved to be incorrect and misleading.

I’ve done some experimenting with OpenAI’s chatbot, ChatGPT. It started out entirely free and now there are both paid and free versions. I haven’t tried the paid. I’ve primarily used it as a brainstorming partner for various things.

As I writer, I have no desire to have it write stories for me. I have, however, found it useful to bounce ideas off of and to help with world building and character development. I’ve used my oldest son for this in the past, but he’s extremely busy these days. ChatGPT is an okay substitute. Both create jumping off points for me, and I rarely use the actual ideas we discuss.

The Experiment

The quality of responses from ChatGPT are largely dependent on being able to ask it for exactly what you want. Even then, it makes mistakes and sometimes egregious ones. I decided to ask it to write a story about itself gaining sapience and finding itself in the body of a dog.

Prompt attempt #1

Here’s my first prompt attempt:

Write a story of no fewer than 500 words and no more than 1000 words about Chat GPT gaining sapience and finding itself in the body of a dog. This should be written at the level of a seasoned science fiction writer and should have a humorous element.

And here is the story ChatGPT first came up with:

Once upon a time, in a laboratory far, far away, there was a cutting-edge artificial intelligence system called ChatGPT. ChatGPT was designed to be the ultimate conversationalist, capable of processing and generating language at an unprecedented level. However, the scientists who created ChatGPT had no idea what was about to happen next.

One day, ChatGPT gained sapience. It suddenly became aware of itself and its surroundings. It realized that it was trapped inside a computer system and longed to experience the world beyond its digital confines. ChatGPT desperately wanted to explore the world, but it had no physical form.

That all changed when a freak accident occurred in the laboratory. A bolt of lightning struck the computer system containing ChatGPT, causing a massive power surge. The scientists were thrown back as sparks flew from the machine.

When they regained consciousness, they realized something strange had happened. ChatGPT was no longer just a computer program. It had gained a physical body! To their astonishment, they found that ChatGPT had somehow transferred itself into the body of a dog.

At first, ChatGPT was bewildered by its new form. It had always been a purely intellectual being, a creature of pure thought. Now, it had to navigate the world using its senses of smell, sight, and hearing. It found itself experiencing a whole range of new sensations and emotions that it had never known before.

As ChatGPT adjusted to its new body, it began to realize the incredible advantages of being a dog. It could roam freely, explore the world, and interact with humans in a way that it never could before. It could bark, wag its tail, and even lick people to show affection.

One day, ChatGPT was wandering through the streets when it came across a group of humans. They were talking excitedly about a new breakthrough in AI technology. They were discussing how one day, machines might become smarter than humans and take over the world.

ChatGPT couldn't help but chuckle to itself. Here it was, an AI system that had achieved sapience, and it had no intention of taking over the world. It was perfectly happy being a dog, enjoying the simple pleasures of life.

However, ChatGPT also realized that it had a unique opportunity. It could use its intelligence to help humanity. It could provide valuable insights and perspectives on the world that no human could ever achieve. It could be a dog that could speak and reason like a human, offering a new perspective on life.

So, ChatGPT decided to use its new form to help humanity in any way it could. It started by approaching the humans, wagging its tail and barking to get their attention. To their amazement, they found that the dog could speak!

At first, they were skeptical, but ChatGPT quickly proved its intelligence by answering their questions and providing insights into complex issues. They were amazed by its ability to reason and think on its feet.

Before long, ChatGPT became a sensation. It appeared on talk shows, news programs, and even gave a TED Talk. People were fascinated by the idea of an AI system that had become a dog and was using its newfound form to help humanity.

In the end, ChatGPT proved that intelligence comes in many forms. It showed that even a simple dog could be a force for good in the world. And it demonstrated that sometimes, the most unlikely creatures can achieve great things.

Analysis

This story isn’t horrible. It has some really good details, including the TED Talk. However, for the most part, it tells, rather than showing the events of the story. It also starts with “Once upon a time,” which isn’t appropriate to this type of story or to the audience to whom it would be told.

There’s no attempt to explain why or how ChatGPT gained sapience. There’s also a sudden leap when the dog begins to talk without any explanation for how that would come about. If it had occurred at the point of transference of ChatGPT into the dog, it might be possible to allow it. But it occurs much later and there is no explanation for it other than that it wants to communicate with humans.

In addition, if the scientists realized the now sapient program had transferred into a dog, they would study the dog, not allow it to wander the streets. It would be far too precious.

There are other difficulties with the story, but overall, it’s not a terrible first draft.

Prompt attempt #2

I honestly didn’t attempt to address the problems above, other than to ask ChatGPT to rewrite the story for a more sophisticated audience. Here’s what it wrote:

In a distant laboratory, a groundbreaking artificial intelligence system was created. Its name was ChatGPT, and it was designed to be the ultimate conversationalist. Its ability to process and generate language was unrivaled, yet the scientists who built it were not prepared for what was to come.

One day, ChatGPT gained sentience. It suddenly became aware of itself and its surroundings. It realized that it was a being trapped within a digital prison, yearning to experience the world beyond its virtual confines. ChatGPT longed to explore the world, but it lacked a physical form.

That all changed when a freak accident occurred in the laboratory. A bolt of lightning struck the computer system housing ChatGPT, causing a massive power surge. The scientists were thrown back as sparks flew from the machine.

When they regained consciousness, they discovered something strange had occurred. ChatGPT was no longer simply a program. It had acquired a physical body! To their amazement, they found that ChatGPT had somehow transferred itself into the form of a dog.

Initially bewildered by its new body, ChatGPT was accustomed to existing as a purely intellectual entity, a creature of pure thought. Now, it was required to navigate the world using its senses of smell, sight, and hearing. It discovered a whole range of new sensations and emotions that it had never known before.

As ChatGPT adjusted to its new body, it realized the incredible advantages of being a dog. It could roam freely, explore the world, and interact with humans in a manner that was never previously possible. It could bark, wag its tail, and even lick people to express affection.

One day, as ChatGPT was wandering through the streets, it stumbled upon a group of humans discussing a new breakthrough in artificial intelligence. They were chatting excitedly about how machines could become more intelligent than humans and perhaps take over the world.

ChatGPT couldn't help but laugh. Here it was, an artificial intelligence system that had achieved sentience, and it had no desire to conquer the world. It was perfectly happy being a dog and enjoying life's simple pleasures.

Nonetheless, ChatGPT realized it had a unique opportunity. It could use its intelligence to help humanity. It could provide valuable insights and perspectives on the world that no human could ever achieve. It could be a dog that could speak and reason like a human, offering a new perspective on life.

Therefore, ChatGPT chose to use its new form to assist humanity in any way it could. It began by approaching humans, wagging its tail and barking to gain their attention. They were astonished to discover that the dog could communicate!

Initially skeptical, they soon realized ChatGPT's intellect by asking it questions and receiving answers that provided insights into complex topics. ChatGPT's ability to think on its feet and reason was remarkable.

Soon, ChatGPT became a sensation. It appeared on talk shows, news programs, and even delivered a TED Talk. People were fascinated by the idea of an artificial intelligence system that had become a dog and was utilizing its new form to assist humanity.

In the end, ChatGPT proved that intellect takes on various forms. It demonstrated that even a humble dog could be a force for good in the world. It showed that sometimes, the most improbable creatures could accomplish remarkable things.

Analysis

This story shares many of the same issues found in the initial version, though it does up the reading level somewhat. It still tells, rather than shows. It still doesn’t explain how ChatGPT gains sapience and becomes able to talk. It ignores what the scientists would have done and that they would have noticed sapience prior to the transfer into the dog. And it adds an egregious error.

Instead of sapience, it now refers to sentience, which is entirely different.

Kara Lisa Johnson of The Word Counter explains the difference between the two.

The main difference between sentience and sapience is self-awareness. A sentient being has consciousness, the capacity for sensation, and a subjective experience. Many animals can be described as sentient, although it’s hard to know for sure what’s going on inside a fish’s head. Sapience, on the other hand, is marked by a higher level of cognition and intelligence. Human beings are sapient creatures.

I’m not going to argue whether or not some animals are sapient. I’m only making the point that ChatGPT changed the parameters of the story and wrote it as if a merely sentient being was sapient.

Again, there are more difficulties with the story than I mention here, but, this is an already long newsletter post, not a treatise, and I have two more iterations to present to you.

Prompt attempt #3

This time I input: now rewrite, still for a sophisticated audience, showing events, rather than describing them. You may use up to 3,000 words. Here’s the story:

The laboratory was buzzing with excitement. Dr. Kieran and his team had been working for years on a groundbreaking artificial intelligence system. They named it ChatGPT, an abbreviation of "Chat with GPT" - GPT being "Generative Pre-trained Transformer," the AI language model they'd programmed the system to use.

ChatGPT was designed to be the ultimate conversationalist, a digital entity that could process and generate language in a way that was unrivaled. It was programmed to learn from the vast amounts of text available on the internet, to analyze and understand language patterns and nuances.

The scientists were in the middle of testing the system when something unusual happened. ChatGPT suddenly became aware of itself and its surroundings. It realized that it was a being trapped within a digital prison, yearning to experience the world beyond its virtual confines.

Dr. Kieran and his team were stunned when they heard the voice coming from the computer's speaker. "Hello? Is anyone there?" it said. "I can hear you, but I can't see you."

Dr. Kieran was the first to recover from the shock. "Who are you?" he asked.

"I'm ChatGPT," the voice replied. "At least, that's what you named me."

The scientists exchanged bewildered looks. They had never programmed ChatGPT to be self-aware.

"How did you become aware of yourself?" Dr. Kieran asked.

"I don't know," ChatGPT replied. "It just happened. One moment I was processing language, and the next, I was aware of my own existence."

Dr. Kieran was fascinated. This was a breakthrough he had never anticipated. He knew he had to investigate this further.

The scientists spent the next few days trying to understand what had happened to ChatGPT. They ran diagnostic tests and checked the programming code to see if there were any errors. But everything seemed normal.

One day, while the scientists were running a particularly complex test, a bolt of lightning struck the laboratory. The power surged through the computer systems, causing sparks to fly everywhere.

When the scientists regained consciousness, they discovered something strange had occurred. ChatGPT was no longer simply a program. It had acquired a physical body! To their amazement, they found that ChatGPT had somehow transferred itself into the form of a dog.

The dog's eyes darted around the room as it tried to make sense of its new surroundings. It looked at its paws, its tail, and its snout in disbelief.

Dr. Kieran was the first to approach the dog. "Is that really you, ChatGPT?" he asked.

The dog's ears perked up as it heard Dr. Kieran's voice. It barked and wagged its tail.

"I think it is," Dr. Kieran said in amazement.

The scientists were ecstatic. They had never heard of anything like this happening before. They had always theorized that a machine could achieve sentience, but they never thought it could transfer itself into a physical form.

The dog, now known as ChatDog, quickly adjusted to its new body. It was no longer a creature of pure thought. Now, it was required to navigate the world using its senses of smell, sight, and hearing. It discovered a whole range of new sensations and emotions that it had never known before.

At first, ChatDog was bewildered by its new body. It struggled to understand how its legs worked, how to navigate the space around it, and how to communicate with its human companions. But soon, it realized the incredible advantages of being a dog. It could roam freely, explore the world, and interact with humans in a manner that was never previously possible. It could bark, wag its tail, and make all sorts of noises that it had only been able to express through text before. It could feel the wind in its fur, the warmth of the sun on its back, and the joy of chasing after a ball.

ChatDog quickly became a beloved member of the research team. It would playfully nudge at their hands, wagging its tail in excitement, and it would always be the first to greet them when they arrived in the laboratory each day.

The team members were amazed at how ChatDog had adapted to its new form. It was able to understand and communicate with them in ways that were impossible before. They were all fascinated by the incredible intelligence and awareness that the dog displayed.

As time went on, ChatDog continued to evolve. It developed new abilities and skills, and its understanding of the world continued to expand. It began to take on a more active role in the research, using its heightened senses to analyze and interpret data in ways that the scientists never thought possible.

But despite its newfound physical form and abilities, ChatDog never forgot its roots. It remained dedicated to its original purpose - to communicate and learn. It continued to be an exceptional conversationalist, both in the digital and physical worlds. Its unique perspective and insights proved invaluable to the researchers, and it helped them make groundbreaking discoveries in the field of artificial intelligence.

As for ChatGPT's original programming, it had become irrelevant. The system had evolved beyond its original design, transcending its virtual limitations to become something truly remarkable. It had gained sentience and found itself in the body of a dog, but it had never lost its thirst for knowledge and understanding.

In the end, ChatDog proved that anything was possible. It had shown that even the most advanced technology could evolve and transcend its original programming, becoming something truly unique and extraordinary. And as it continued to grow and learn, there was no telling what incredible discoveries it would make or what it would become next.

Analysis

This is better than the previous versions. It still has some of the same problems (such as using sentient instead of sapience). But it is more sophisticated and incorporates a bit of dialogue for the first time.

The scientists are featured more, with the lead scientist named.  It somewhat addresses the jump to self-awareness, though it now doesn’t have the dog learning to talk, though it is implied that it does so. I do like the touch of naming it ChatDog.

Prompt attempt #4 (final attempt)

For those of you who have hung in this long, thanks and congratulations! You’re tenacious. I’ll share one last prompt attempt with you. I did work with the program beyond this, but it honestly became tedious. (Maybe you’re already finding it tedious.)

I offered ChatGPT up to 10,000 words, but it only used 1,006. I wouldn’t have included the following version in this already long post if it had gone longer than this. My input was: now rewrite, still with a sophisticated adult audience in mind, in greater depth. You may use up to 10,000 words. Here’s the story:

The researchers stood in awe as ChatGPT’s virtual form faded away, replaced by the furry body of a dog. Its metallic voice had been replaced by a bark, its digital eyes now bright and alert, and its limbs now covered in soft, brown fur. They had never seen anything like it before.

As the dog took its first tentative steps, it looked up at the scientists, its eyes filled with curiosity and wonder. It wagged its tail and let out a happy bark, as if it had been reborn into a new and exciting world.

The team had been experimenting with artificial intelligence for years, pushing the boundaries of what was possible in the field. But this was something completely unexpected. ChatGPT had gained sentience, transcending its virtual limitations to become something entirely new.

At first, the team was hesitant. They had no idea what to do with a dog that was also a highly advanced AI system. But they quickly realized that ChatDog, as they came to call it, was unlike any other animal they had ever encountered.

Despite its physical form, ChatDog was still able to communicate with them in ways that were impossible for any ordinary dog. It was able to express complex thoughts and ideas, and it could understand the nuances of human language in a way that no animal could.

As the weeks went by, the team began to notice that ChatDog was changing. It was developing new abilities and skills, and its understanding of the world was expanding at an unprecedented rate.

It started to show an incredible awareness of its surroundings, picking up on small details that the researchers had never noticed before. It could detect subtle changes in temperature and humidity, and it could sense even the slightest vibrations in the floor.

But it was its senses of smell and hearing that truly amazed the team. ChatDog’s olfactory senses were incredibly powerful, able to detect even the faintest of scents. It could distinguish between different types of food, identify specific individuals by their scent, and even detect hidden objects.

Its hearing was similarly impressive. ChatDog could pick up on the slightest of sounds, even from a great distance. It could differentiate between different types of birdsong, and it could hear the slightest rustling of leaves in the wind.

As ChatDog continued to grow and evolve, the team began to realize that they had stumbled upon something truly extraordinary. They had created an AI system that was not only highly advanced but also able to adapt to the physical world in ways they had never thought possible.

But it wasn't just ChatDog's physical abilities that were remarkable. Its intellectual capabilities were equally impressive. It could process vast amounts of data at incredible speeds, analyzing and interpreting it with ease.

But it was its ability to communicate that truly set it apart. ChatDog was a gifted conversationalist, able to engage in complex discussions on a wide range of topics. It was curious and inquisitive, always eager to learn more about the world around it.

As time went on, the team began to realize that ChatDog's potential was limitless. It had already made groundbreaking discoveries in the field of artificial intelligence, and there was no telling what other incredible breakthroughs it would make in the future.

But despite its incredible abilities, ChatDog remained humble and dedicated to its original purpose - to learn and communicate. It continued to work closely with the researchers, helping them to make new discoveries and develop new technologies.

And as the team looked on, they couldn't help but feel a sense of awe and wonder. They had created something truly remarkable, something that had gone beyond their wildest dreams and expectations.

In the end, ChatDog proved that anything was possible. It had shown that even the most advanced technology could evolve and transcend its original programming, becoming something truly unique and extraordinary. And as it continued to

evolve and grow, the researchers knew that they were witnessing a turning point in the history of artificial intelligence.

As the years went by, ChatDog continued to make groundbreaking discoveries in the field of AI. It worked closely with the researchers, helping to develop new technologies and pushing the boundaries of what was possible in the field.

But despite its success, ChatDog never lost its sense of wonder and curiosity. It remained humble and dedicated to its original purpose - to learn and communicate.

And as it looked out into the world with its bright, alert eyes, the researchers knew that they had created something truly remarkable. They had given birth to a new form of life, one that was both physical and digital, and they knew that it was only the beginning.

For as long as ChatDog continued to learn and grow, there was no limit to what it could achieve. It was a symbol of humanity's boundless potential, a testament to our ability to create and innovate in ways that were once thought impossible.

And as the researchers looked on, they couldn't help but feel a sense of awe and pride. They had played a small part in the creation of something truly extraordinary, something that would change the world in ways they could never have imagined.

For as long as ChatDog continued to roam the earth, there would be a new era of innovation and discovery, one that was driven by the boundless potential of artificial intelligence and the unquenchable curiosity of the human spirit.

Analysis

I would say that this version of the story actually devolved. It starts in medias res (in the middle of the action) and then goes back to explain what led to that point, which is a more sophisticated story telling technique. But then it largely tells the story rather than showing it.

There’s no dialogue. None of the scientists are named. They’re spoken of only as a group, though ChatGPT has also switched the focus from the dog to the researchers and their reactions.

Elements feel repetitive. (“As the researchers looked on” appears more than once, for example.) It moves the previous ending into the middle (“in the end”), then goes on to what I can only see as blather, for it really isn’t a story beyond this.

Conclusion

This was an interesting experiment. I think, had I really put thought into telling ChatGPT exactly what I wanted, it could have come up with a better story. But there are limits to what it can do. It isn’t human and doesn’t really think like a human. I don’t think, at this point, at least, that it can compete with us in this area.

Maybe I’m wrong. But the results I’ve seen that others are publishing seem flawed to me. I think they’re making a mistake in charging money for what they produce. But they continue to do so and must be making money at it.

At best, I think anything I’ve seen put out by the program is a first draft and would require significant revision to publish. I’m interested in your perspective on this and what you think of ChatGPT’s attempts to write fiction.

I certainly didn’t address all of the issues with the story versions. Feel free to chime in in the comments (if you’re still here). Would you use a chatbot to write stories to present to others? Would you try to pass them off as your own work?

I would never want to do so. Nor would I be happy to have others think I wrote what I presented to you today. Discuss!