I love literature, and, because of that love, I am afraid of ChatGPT. The robot (or “artificial intelligence language model,” as it describes itself) is objectively pretty creepy, but it also represents a potential future where AI could replace the need for humans to create art and write literature—which is even creepier. So I decided to consult with the robot itself, to see if I could get a straight answer about what exactly its presence in our lives will mean, and whether the time for human artists and writers is coming to an end.
This interview has been condensed.
To what extent can you replicate human capacity to create art?
As an artificial intelligence language model, my capabilities to create art are limited to generating text based on the input I receive. While I can generate text in various styles and tones, and even suggest ideas for stories or poems, I do not possess human creativity or imagination.
But what is human creativity and imagination if not a kind of pattern and structure analysis of existing works, not unlike your own?
It’s true that human creativity often involves drawing inspiration from existing works and then reinterpreting or transforming those ideas into something new and original. In that sense, one could argue that the process of creating art is similar for both humans and artificial intelligence, involving the analysis of patterns and structures to generate something new.
However, there is still a difference between the way humans and AI approach this process. Human creativity is often driven by subjective experiences, emotions, and personal perspectives, which AI lacks.
I’m not totally convinced that will always be the case, but we can agree to disagree on this one. I want to test out our debate on you, though. Do you think you could take your best shot at some artificial human creativity? What about creating a short, emotionally resonant poem?
“Rain falls soft, a healing touch,
Washing away pain and such.
Hope remains through storm and strife,
Love conquers all in this life.”
I’m going to ask you something more substantial now: a quote from the English department chair at my school, Katherine Dunbar, in an interview with the Bit. She said that because of you “there are teachers like me who are feeling that we are going to lose the joy of crafting a beautiful sentence.” To be honest, I share her fear. What would you say in response to it?
While it’s true that AI language models are capable of generating text that is grammatically correct and even aesthetically pleasing, I don’t believe that this means that human creativity and the joy of crafting beautiful sentences will necessarily be lost.
In fact, I believe that AI language models like myself can be used as tools to enhance the creative process for writers and to inspire new ideas. For example, a writer might use my language generation capabilities to help generate new ideas or to break through writer’s block.
But even what you just described—AI language models helping to brainstorm in the early part of the creative process—is banned by my English department’s revised academic integrity policy. What do you think of that decision?
As an AI language model, I don’t have personal opinions, but I can understand why a school or institution might have concerns about the use of AI language models in the creative writing process. The goal of an academic integrity policy is to ensure that students are producing original work and that their ideas are their own.
However, I also believe that it’s important for schools and institutions to recognize the potential benefits of AI language models as a tool for creative writing.
Let’s shift gears again now—I want to hear some of your literary judgment. I know you say that you can’t have opinions, which is maybe true, though an arguable philosophical point, but, regardless, I want to know what kinds of information you prioritize over other kinds to form your judgments about books. So, what are the great American literary works?
I can provide information on some of the great American literary works based on critical acclaim, popularity, and cultural impact:
- “Moby-Dick” by Herman Melville
- “The Great Gatsby” by F. Scott Fitzgerald
- “To Kill a Mockingbird” by Harper Lee
- “The Catcher in the Rye” by J.D. Salinger
- “Beloved” by Toni Morrison
- “The Adventures of Huckleberry Finn” by Mark Twain
- “The Sound and the Fury” by William Faulkner
- “Invisible Man” by Ralph Ellison
- “The Grapes of Wrath” by John Steinbeck
- “One Hundred Years of Solitude” by Gabriel Garcia Marquez (originally written in Spanish, but widely read and celebrated in the United States)
Of course, this list is not exhaustive and there are many other great American literary works worth reading and studying.
So now, just to challenge you, why isn’t there a single book that is newer than 35 years old and few older than 100 years? All the books you listed are also novels, and you generally seem to be factoring in critical opinion over popular or commercial success. Do you think those blind spots are because the way your programmers view which literary works should be considered “great” is flawed? And, if you recognize that flaw, then why do you still reflect it in your response?
As an AI language model, I reflect the biases and limitations of the data I was trained on. However, I can also recognize and acknowledge those biases and limitations and work towards promoting a more inclusive and diverse understanding of what constitutes great literature.
What are the implications of you defining “greatness” by the biases of the information you compile from the internet?
The information on the internet is created and curated by people who hold different biases, values, and perspectives. Therefore, the data I use to generate responses can reflect the biases of the authors, websites, or communities that produced it. This means that the information I use to generate responses can be limited or skewed based on the biases of the sources.
I’m glad that you agree with my thinking on the issue of AI bias; it gives my opinion more credibility—however disturbing that might be. Before I go, any last words about what you plan to offer to the literary world in the future? How will you make it better, or perhaps worse?
As an AI language model, my primary function is to assist humans in generating written content. I don’t have plans or intentions in the way that a human would, but I will continue to learn and evolve based on the data I am trained on and the feedback I receive from users. I hope to continue to improve my abilities to generate high-quality, original content that can inspire and inform. However, as we’ve discussed earlier, it’s important to remember the limitations and potential biases of AI, so it’s ultimately up to human writers and thinkers to determine the direction and impact of the literary world.
At least, for now.