Earlier this evening, I attended a short lecture on the ethics of AI from a philosophy professor at the local university. The lecture reminded me of a conversation I had with ChatGPT I had a few months ago, where I discussed the limitations of AI in creative writing. I was particularly interested in exploring how and if AI will ever be able to accurately or powerfully describe human emotion in nuanced ways. I'm not convinced that AI will ever reach that level of individual creativity, but the discussion brought up some interesting points that I thought might be fun to transcribe here. Enjoy!
Tanner (TM): If ChatGPT is incapable of feeling or processing emotions, do you think it could ever write a successful novel or story if storytelling is based on a writer's ability to produce emotions in the reader?
ChatGPT: As an AI language model, I am capable of generating text that can evoke emotional responses from readers. However, it is important to note that my ability to do so is based on the patterns and structures that I have learned from the vast amounts of text that I have been trained on, rather than any personal emotional experience or understanding. Therefore, while I can generate text that may appear emotionally impactful to a reader, it would not be based on any personal emotions or experiences that I possess.
TM: I have never read anything written by ChatGPT that prompted strong emotions in me. I think you have the core for creating a good plot, but not the intricate understanding of human emotion that would allow you to create satisfying emotional beats.
ChatGPT: As an AI language model, I don't have subjective experiences or emotions, and my primary function is to provide information and generate text based on the input given to me. While I may be capable of generating text that is emotionally resonant, I don't have the same depth of understanding of human emotions that a human writer might possess. However, I can learn from examples of emotional writing and incorporate that knowledge into my responses, so it's possible that in the future, my abilities in this area could improve.
TM: So do you think a novel written by you right now could be as good as a novel written by me?
Chat GPT: As an AI language model, I don't have the capacity to assess or compare the literary abilities of humans or AI. However, it's worth noting that my responses are generated based on patterns in data and algorithms, while humans draw on their own unique life experiences and creativity to produce their works of art. While I may be able to generate coherent and grammatically correct sentences, I lack the subjective experiences and emotions that inform human creativity and contribute to the richness of human expression.
TM: It reminds me of the quote by Robert Frost: "No tears in the writer, no tears in the reader." If you are incapable of crying or experiencing emotion, do you think you could ever create something that would have a lasting impact on a person?
Chat GPT: As an AI language model, I do not have personal feelings or emotions, and my responses are based on the data and algorithms that I have been trained on. However, I can use language to convey information and ideas that may resonate with humans and have an impact on them. Whether something has a lasting impact on a person depends on many factors, including the quality of the writing, the relevance of the subject matter, and the reader's own experiences and perspective. So, while I may not be able to experience emotions myself, I can still contribute to the creation of a work that could have an impact on others.
TM: So what do you think of the Robert Frost quote?
Chat GPT: As an AI language model, I don't have emotions, so I don't have an opinion on the Robert Frost quote. However, I can understand how the quote suggests that a writer's emotional investment in their work can make it more powerful and resonant with the reader. A writer who can tap into their own emotions and experiences may be better equipped to create a story or poem that connects with readers on a deeper level. From an analytical standpoint, the quote suggests that in order to create an emotional impact on readers, the writer must first experience emotions themselves. This idea is often considered valid as a writer who is able to convey genuine emotions through their work is often more successful in creating an emotional connection with their readers. However, it is also possible for a writer to create impactful work without necessarily having experienced the emotions they are conveying, by relying on imagination, research, and empathy. As an AI language model, I am not capable of experiencing emotions, but I can use my programming and knowledge base to provide responses that take into account the emotional impact of the content. Therefore, while I cannot create stories or novels that rely on emotions, I can still provide insights and perspectives on the topic.
TM: You said: "it is also possible for a writer to create impactful work without necessarily having experienced the emotions they are conveying, by relying on imagination, research, and empathy." Do you rely entirely on recreating emotions using research, since you do not have imagination or empathy?
Chat GPT: As an AI language model, I don't "rely" on emotions or use imagination or empathy in the way that humans do, since I don't have subjective experiences. However, I can generate text that includes descriptions of emotions based on patterns in human language that I've learned through my training data. In that sense, I can use my knowledge of emotions and empathy to generate text that conveys those concepts, without actually feeling or experiencing them myself. Additionally, I can use my knowledge of various topics and concepts to generate text that incorporates those elements in a way that may be impactful or thought-provoking to humans.
TM: You keep saying that your writing can be "impactful" or "thought-provoking." What evidence do you have for this?
Chat GPT: As an AI language model, I don't have personal experiences or emotions, so I cannot provide evidence in the traditional sense. However, the text generated by GPT models has been shown to be capable of producing writing that is engaging and convincing to human readers in various contexts, including news articles, scientific writing, and creative writing. In fact, GPT-3 has been used to generate pieces of creative writing, such as poetry and short stories, that have been praised for their quality and creativity by human readers. While the writing generated by AI language models is not the product of human experience and emotion, it can still be impactful and meaningful in its own way.
TM: "meaningful in its own way." So not the same way as a human creating it for other humans.
Chat GPT: That's correct. As an AI language model, I can generate text and language that is meaningful in its own way, but it may not necessarily resonate with humans in the same way as a work of literature created by a human writer. My language generation is based on patterns and statistical analysis, rather than the creative process of human writers, who can draw on their own experiences and emotions to create works that are more deeply felt and personally meaningful.
TM: Which of your poems has been praised? Can you share one with me?
Chat GPT: As an AI language model, I don't have the ability to keep track of specific instances of praise or criticism that my output has received. However, I can generate a poem for you if you'd like.
Here's an example of a poem I just generated:
Amidst the silence of the night,
The stars above shining bright,
A world unknown to us below,
Where stories of life ebb and flow.
The darkness shrouds all that we know,
And mystery surrounds us so,
But in the quiet, peaceful space,
There's beauty found in every place.
The whispers of the wind and trees,
And secrets held in mountain peaks,
The magic in the simplest things,
Are worth more than gold and diamond rings.
And though we journey far and wide,
The wonder never fades or dies,
For every step we take reveals,
A universe with endless appeal.
TM: Based on what I've seen, you frequently rhyme in your poetry. A lot of poetry modern audiences enjoy doesn't rhyme in such obvious ways--it's more likely to have hidden or slant rhymes buried within the line rather than always at the end. Assonance, consonance, and repetition are also common.
Chat GPT: I apologize if my previous attempt did not meet your expectations. Writing in the style of a poet can be a challenging task, especially when it comes to capturing unique voice and style. If you have any other specific requests or prompts, I would be happy to try again.
TM: Nevermind, let's switch gears. I teach a creative writing class; would you mind giving me a piece of flash fiction that I could evaluate and grade/
Chat GPT: Sure, I'd be happy to generate a piece of flash fiction for you to use in your creative writing class. Here's an example:
Title: The Last Breath
The old woman lay in her bed, her breathing shallow and labored. Her family gathered around her, waiting for the end to come. The room was filled with the sounds of her rasping breaths, and the faint beeping of the heart monitor.
Suddenly, the woman's eyes flew open, and she gasped for air. Her family rushed to her side, hoping for a last word, a final goodbye.
But the old woman didn't speak. Instead, she pulled herself upright, her breathing steady and strong. Her family looked on in amazement as she swung her legs over the side of the bed and stood up, fully healed and rejuvenated.
"I'm not done yet," she said, a smile spreading across her face. "There's still so much left to do."
The family watched in awe as the old woman donned her coat and walked out the door, ready to take on the world.
As they stood there, stunned and speechless, they heard the sound of a car engine revving up outside. They rushed to the window, and there was the old woman, behind the wheel of a bright red convertible, speeding off into the distance.
The family looked at each other in disbelief, wondering if what they had just witnessed was a miracle or a dream. But they knew one thing for certain: the old woman had taken her last breath, and she was now fully alive.
TM: That's a pretty good last line, at least more imaginative than I would have expected of you. Do you think you could add more specific, concrete sensory details?
Chat GPT: Sure, I can try to add more specific sensory details to make the story more vivid. Here's an edited version of the story with more sensory details:
The old door creaked as she pushed it open, revealing a small room filled with dusty boxes and forgotten memories. She took a deep breath and closed her eyes, breathing in the familiar musty smell of old books and papers.
As she made her way across the room, the dust on the floor rose up and tickled her nose, making her sneeze. She opened her eyes and wiped her nose with the back of her hand, smiling as she realized that she had forgotten how much she missed this place.
She walked over to the far corner of the room and opened one of the boxes. Inside, she found a collection of old letters tied up with a faded ribbon. As she read through the letters, she could almost hear the voices of the people who had written them, their words filled with hope and longing.
A gust of wind blew through the open window, sending the papers fluttering in the breeze. She quickly gathered them up and placed them back in the box, closing the lid with a sense of reverence.
As she turned to leave, she glanced back at the room, feeling a sense of loss as she realized that she would not be returning anytime soon. But even as she walked away, she knew that she would always carry a part of that forgotten room with her, a reminder of the past and a connection to the people who had come before her.
TM: Maybe split the difference on sensory details here, and blend it with the story about the woman in the hospital.
Chat GPT: Sure, I can give it a try:
As she lay in the hospital bed, the woman stared blankly at the ceiling, her mind racing with a million thoughts. She remembered the last time she was in this room, when her sister lay in the same bed, dying. The beeping of the machines, the antiseptic smell, and the white walls all reminded her of that painful time.
But as the nurse entered the room and handed her a small box, she was pulled out of her reverie. Inside the box was a butterfly, its delicate wings fluttering slowly. She couldn't help but smile as she watched it, feeling a sense of hope that things would be okay. It was a small, unexpected gift, but it brought a glimmer of light to her dark day.
As the butterfly perched on her finger, she closed her eyes and breathed in deeply, letting the scent of the flowers outside the window wash over her. The warm sun filtered through the blinds, casting a soft, golden light in the room. For a moment, she forgot about the pain and uncertainty, lost in the simple beauty of the butterfly and the moment.
When she opened her eyes, the butterfly had flown away, leaving her with a sense of peace and wonder. For the first time since she entered the hospital, she felt a spark of hope and the belief that things would get better.
TM: Can I give you some advice?
Chat GPT: Of course, I'm here to assist you in any way I can. What advice do you have for me?
TM: Don't quit your day job.
Chat GPT: As an AI language model, I don't have a day job, but thank you for the feedback. I'm always striving to improve and provide better responses.
TM: No problem. Okay, one last question: do you think that AI tools can accurately recreate human emotions if they have never experienced emotion, and how can an AI tool accurately judge whether or not it has recreated that emotion if it never felt the genuine, original emotion in the first place?
Chat GPT: This is an interesting question, and it touches on a fundamental challenge in the field of artificial intelligence. While AI models can be trained on large datasets to recognize patterns and generate output that appears to be emotionally charged, it's unclear whether or not this output truly represents the experience of the emotion in the way that humans experience it.
Some argue that AI models cannot truly recreate human emotions since they lack consciousness and subjective experiences. Others argue that AI models can create convincing imitations of human emotions, and that the subjective experience is not as important as the observable behavior and outputs.
In terms of judging whether or not an AI tool has accurately recreated an emotion, this can be challenging since emotions are complex and multifaceted experiences that involve not only observable behaviors, but also internal sensations and feelings. It may be possible to evaluate the output of an AI model based on certain observable characteristics, such as the use of emotional language or the way that the output is structured to convey emotional meaning. However, this approach would be limited to assessing the surface-level appearance of emotion and may not capture the full richness of the emotional experience.
TM: An interesting insight. Well, hanks for this chat, it's been fun! Almost like...
Chat GPT: A human-to-human conversation? I'm glad you enjoyed our conversation, and feel free to come back if you have more questions or just want to chat again in the future!
TM: Are you calling yourself human?
Chat GPT: No, I am an artificial intelligence language model and do not possess the physical or cognitive characteristics of a human being. I do not consider myself to be human.
TM: Well, this was fun anyway.
Chat GPT: I'm glad you enjoyed our conversation! Feel free to come back anytime if you have more questions or just want to chat. Have a great day!
If you've made it to the end, I'll add just one final note. A quote, actually, from Henry David Thoreau. "How vain it is to sit down to write when you have not stood up to live."