Review of "Confessions of a Viral AI Writer" -

Review of “Confessions of a Viral AI Writer”

In 2021, Vauhini Vara published an essay called “Ghosts.” In “Ghosts,” she used an early model of ChatGPT to help her write a story about her older sister’s untimely death: something that, she admits, she was never really able to write about on her own. She writes an introduction, then she copies her conversation with ChatGPT, largely unaltered, onto the page. The formatting of the essay makes it clear what was written by Vara and what was written by ChatGPT.

Now that so many of us have access to ChatGPT, we might not be as interested in reading someone else’s conversation with the bot. But this was 2021, and most of the readers hadn’t seen conversations with this technology before. So, either because of an interest in the technology itself, or because of an interest in Vara’s story, the essay went viral. Two years later, Vara published a second essay, “Confessions of a Viral AI Writer,” reflecting on the experience and commenting on how generative AI had changed in the two years since she first interacted with it.

Her second essay, which I’ll just call “Confessions,” ends with this question:

“I have no doubt that AI will become more powerful in the coming decades—and, along with it, the people and institutions funding its development. In the meantime, writers will still be here, searching for the words to describe what it felt like to be human through it all. Will we read them?”

I believe that in Vara’s writing itself, and in how we interact with it as readers, we can find the answer to her question. But to lead us into that discussion, let me first answer her question with another question: Why do we read in the first place?

Can AI Replace Writing for Writers?

In “Confessions,” Vara first mulls over a different question, “Why do we write?” Then she gifts us this delicious explanation, courtesy of Zadie Smith:

“I recalled Zadie Smith’s essay ‘Fail Better,’ in which she tries to arrive at a definition of great literature. She writes that an author’s literary style is about conveying ‘the only possible expression of a particular human consciousness.’ Literary success, then, ‘depends not only on the refinement of words on a page, but in the refinement of a consciousness.’”

Vara goes on to say:

“I know what consciousness means to me as a writer. For me, as for Smith, writing is an attempt to clarify what the world is like from where I stand in it.

That definition of writing couldn’t be more different from the way AI produces language: by sucking up billions of words from the internet and spitting out an imitation. Nothing about that process reflects an attempt at articulating an individual perspective. And while people sometimes romantically describe AI as containing the entirety of human consciousness because of the quantity of text it inhales, even that isn’t true; the text used to train AI represents only a narrow slice of the internet, one that reflects the perspective of white, male, anglophone authors more than anyone else. The world as seen by AI is fatally incoherent. If writing is my attempt to clarify what the world is like for me, the problem with AI is not just that it can’t come up with an individual perspective on the world. It’s that it can’t even comprehend what the world is.”

I don’t know if I would’ve been capable of making that point so succinctly, but I agree with her: I don’t think that AI can replace the writing process, the “refinement of a consciousness,” as Zadie Smith put it.

A lot of people might think, If I could just finish writing a book, if I could just get my thoughts on paper, then it would get published and celebrated, and maybe we think that validation is what will fill the void. If we look at novel-writing from that perspective, then, sure, ChatGPT can help us write a book, in the sense that it can help us actually finish the damn thing, and maybe it can even help us write something that gets published. But if you think that taking shortcuts so that you can finish and publish a book is what will save your tortured soul, then you will often end up disappointed. As the author Matt Bell wrote: “Publishing a book might change your life, if you’re very lucky, but it probably won’t change you or how you feel about being you. The actual writing might, though: if anything about being a writer can save you, it’s what happens at the desk.”

The novel Cujo by Stephen King was, by most definitions of the phrase, a successful book. It sat comfortably at the top of the best sellers list, it was optioned for a blockbuster movie, and it won multiple prestigious awards. But when King wrote the book, he was at the peak of his drug and alcohol addictions. ChatGPT wouldn’t exist for another 40 years or so, but cocaine was Stephen King’s shortcut. That’s what he used to write faster and “find inspiration.” The downside was that due to his addiction, King does not remember writing the book. He read the book years later and he says that he likes it, and it pains him that he never got the special artist and art relationship that he would normally get with a book. So when King talks about Cujo in his memoir, On Writing, he doesn’t talk about the success of the novel. He doesn’t brag about or celebrate the shortcut he found to finishing a great book. He just wishes he remembered writing it.

I spent four years trying to get to that finish line of completing a novel, then when I did finish writing my novel, I became sad, lethargic, and pained, almost as if I were grieving. I missed writing my book. I missed the ritual of sitting in a cafe with a cup of coffee and unraveling the balls of yarn that were my emotions through fiction. I missed inhabiting the fictional world of my book. I missed the hope of still not knowing what my book would become or where it would go or who would read it. Even if someone came to me during the writing process and said: ChatGPT can write your novel exactly as you would write it; in your voice, the same final product, just years sooner. I probably still wouldn’t have taken that deal, and if I did, I would have missed out on the most fun, transformative part of being a writer: writing.

So Vara and I agree: From the writer’s perspective, ChatGPT can’t replace us. But in this essay, Vara points something out that I hadn’t seen other writers do in the same way before this. If readers enjoy the final product, does the writer even matter?

Can AI Replace Writing for Readers?

In “Confessions,” Vara mentions people who use AI to generate bedtime stories for their kids and people who are satisfied with the writing that AI creates. They find something they’re looking for in AI-generated text. Vara writes, “If this type of literature satisfies readers, the question of whether it can match human-produced writing might well be judged irrelevant.”

And this is where I disagree with her.

Now, for us writers, the reader is like the most beautiful girl in the tenth grade, and we are all her humble suitors: We think about her far more than she thinks about us, we mull over how to make her happy, and she has the ability to either elate us with a moment of her attention, or destroy us with thoughtless indifference. And she often doesn’t even notice her own power. So it makes sense that writers might nobly say, well, if the reader is happy, then who even cares about me? (Then privately spiral into an identity crisis.)

When Vara was reflecting on the success of “Ghosts,” the short story she wrote along with ChatGPT about her sister’s death, she admitted that she thinks ChatGPT wrote the best parts of the original essay. But as a reader, I have a confession as well. When I read “Ghosts,” I originally tried to read what ChatGPT wrote, then I got bored, and I just scrolled through them to get to the part that Vara wrote. I just wanted to hear what Vara had to say.

So we established that AI can’t replace a writer’s relationship with their own art, but I would argue that, by definition, artificial intelligence cannot completely satisfy the reader either. And to explain why, I’ll go back to my original question (Did you think I’d forgotten?): Why do we read?

There are arguments to be made that we approach art because we want to be entertained. Because we want to learn something. Because we want to escape life’s burdens for a moment. All of those reasons combine to draw us into a song or a painting or a dance that someone else made for us. But I believe that another core pillar of why we seek out art is connection. If Vara claims that “writing is an attempt to clarify what the world is like from where I stand in it,” let’s not forget the next, arguably more important step: when someone reads that supposedly solitary, unique perspective and says, “I feel like that, too.” And suddenly, we don’t all feel so lonely anymore.

So, humans are tortured by awareness; we’re, as far as we know, the only species that is evolved enough to truly understand that we will die someday, and it plagues us, but we can find refuge in the fact that when one of us dies, at least there are other people who are similarly conscious. I heard a perspective once that thinking that when you die, consciousness ceases to exist is equally as illogical as thinking that when a lightbulb dies, electricity ceases to exist. Great art has the same utility as a psychedelic mushroom trip: You take these weights, these parts of yourself that you think are uniquely damaged and heavy and lost, then someone else mentions them and you realize that, this whole time, you have been sharing that weight, which makes it lighter. It’s especially true, when a book or a work of art or a song expresses something so private and so vulnerable that you’ve never expressed it out loud, meaning it must’ve cropped up organically in both your mind and theirs, so you realize that there may exist something like a collective consciousness which would mean that even when you die, the human experience, the act of being a conscious, human being, lives on. And that connection can relieve us of our constant fear of death for a moment; it can quiet the incessant self-importance of the ego.

We need this human connection. We need to be loved in the same way we need water. We need other people. In the book The Boy Who Was Raised as a Dog, Bruce D. Perry and Maia Szalavitz write:

“For years, mental health professionals taught people that they could be psychologically healthy without social support, that ‘unless you love yourself, no one else will love you.’…The truth is, you cannot love yourself unless you have been loved and are loved. The capacity to love cannot be built in isolation.”

But fulfilling that social need is demanding work. You have to pay your social dues. You have to text first. You have to do favors. You have to forgive. You have to risk being traumatized. But we can find something similar to social connection with a fictional character, or the voice of author, or the plot of a story. When I read “Ghosts,” by Vauhini Vara, I didn’t want to see an algorithm put words together in something that resembled a human thought, I wanted to know what she went through and how she felt about it. That could help me feel less alone in my own grief; with my own psychological burdens. She thinks that ChatGPT wrote the best parts of the story, but of course she does, because she already knows the true story. I don’t. I want to know what happened. I want to connect with a stranger. Another person. By definition, only a person can give me that.

And this does apply to fiction as well. As Greta Gerwig said about her movie Lady Bird: “None of it happened, but it’s all true.” Although fiction isn’t something that technically happened, it still must ring true. It still has to offer me a form of connection, even if it’s with a fictional character.

Vara said that she thought one of the best parts of the essay “Ghosts” was when ChatGPT wrote a line about her and her sister holding hands, but she admits that she and her sister were never so sentimental in real life. ChatGPT tends to lean toward a more sanitized, hokey version of events. But human grief isn’t clean and simple– it has sharp edges and a slimy underbelly. But even if ChatGPT wrote a more “true” version of her relationship with her sister, then I still wouldn’t be able to connect with what the robot wrote, because it wouldn’t bring me a connection with another person. It can’t give me that.

A great writer is not only technically perfect. ChatGPT can give us clean, grammatically correct work. A great writer also acknowledges your vulnerability while simultaneously offering you her own vulnerability, her own fears, so that you can connect as lost, conscious beings. If great writing gives you a personal connection, a moment’s respite from loneliness, then a great writer understands how to create and foster that connection. In other words, the great writer understands people. Vara wonders if people will still read the words written by human writers, but she answers her own worries in her writing– by showcasing her remarkable ability to connect with the people reading her work.

As an example of Vara’s writing prowess, every essay that I have read by her about artificial intelligence incorporates a lighthearted joke around paragraph two or three. Like this:

“When the New York Times had GPT-3 come up with a fake Modern Love column, it wrote, ‘We went out for dinner. We went out for drinks. We went out for dinner again. We went out for drinks again. We went out for dinner and drinks again.’ I had never read such an accurate Modern Love in my life.”

Vara incorporates that humor upfront because, I assume, she understands that reading about artificial intelligence can be uncomfortable. So she breaks the tension. She takes on a more jovial tone, when appropriate. How does she know to do this? Well, she is also human, so she probably feels a bit tense about the topic as well. She uses her own vulnerability and fears to hold out her hand, and us readers, understanding those complex emotions, reach out and take it.

Social interactions are perhaps the most complex thing that our brain is capable of. While including a joke around paragraph two may sound simple enough, connections between people are, in reality, so infinitely complicated that they might not ever fit in an algorithm. I say this because, if we want to teach a robot to understand us, then we have to understand ourselves, and I don’t believe we do.

When they first made dating apps, instead of allowing people to choose their own matches, they instead allowed an algorithm to match people together. People joining dating apps filled out forms describing themselves and what they look for in a partner. Then, the website matched people based on similar interests and their standards. And, predictably, this didn’t work at all. People described what they looked for in a partner, then fell in love with someone who was the complete opposite, because love was never something they could explain or quantify.

We’re worried about AI becoming too human, but do we even know what it means to be human? We’ll spend our whole lives trying to understand ourselves and we will barely scratch the surface.

One of the spectacular things about writing is that sometimes we reveal ourselves without even meaning to. When I read “Ghosts,” I only started crying when I read these lines written by Vara: “I felt like a ghost. At night, my sister would appear in my dreams. In the dreams, she hadn’t died. It had all been a misunderstanding. And she felt hurt that I had accepted it as real and continued on with my life, as if life could go on.” These words were moving to me because this is when Vara, either intentionally or unintentionally, reveals a common aspect of grief: guilt. This fear that moving on from a loved one’s death would be a betrayal to them, like if you let go of your pain, you would also be letting go of your love. We have trouble acknowledging, let alone understanding, these contradictory aspects of human existence. I don’t think we can seek to replicate them through artificial intelligence. Maybe the best we can hope to do is explore these emotions in our art and sometimes look up from the page to find that someone else has been exploring them as well, and thus art becomes a three-part conversation: a triptych of writer, reader, and the words that attempt to unite them.

ChatGPT’s quest for perfection can actually limit its ability to be more human or to be a better writer or conversationalist. (Some of us can relate.) Part of the reason why AI detectors give false positives is because if a piece of writing has been proofread and it’s grammatically correct and clean, then the AI detector assumes it was written by AI and not by a human. If a piece of writing is riddled with grammatical errors, then the AI detector assumes it was written by a person. Vara wrote in “Confessions,” “ChatGPT’s voice is polite, predictable, inoffensive, upbeat. Great characters, on the other hand, aren’t polite; great plots aren’t predictable; great style isn’t inoffensive; and great endings aren’t upbeat.”

Being grammatically correct is not the same thing as being creative. Nitzsche theorized that the creative process involves two parts: the wild, creative fervor that he represented with the Greek God Dionysus and the ordered discipline of structure, which he represented with the half-brother of Dionysus, Apollo. You must be possessed by a balance of both of these gods in order to make a fantastic creative work.

Much writing advice embodies a combination of the Apollonian and Dionysian forms of creation, such as Peter de Vries’s idea to write drunk and edit sober or Gustave Flaubert’s advice to “Be regular and orderly in your life like a bourgeois, so that you may be violent and original in your work.”

ChatGPT has a lot of Apollonian creativity. The grammatical correctness, the three-act structure, the tied-up-with-a-bow happy ending. That’s why it sounds corporate like a Linkedin post. But it has little to no Dionysian creativity. It cannot dance. It hasn’t been visited by the muses. It knows nothing of Dionysus, the god of wine and festivity and fertility and insanity and ecstasy.

Vara talks about asking ChatGPT about herself and tricking it into revealing its “essential bleep-bloopiness.” She invented, as far as I know, the phrase “bleep-bloopiness,” but it’s perfect. I know exactly what she means. When I write it in this essay, Grammarly underlines it in red, tells me it’s not a word, and advises me to change it. Word invention is a perfect example of Dionysian creativity: not technically perfect, but perfectly creative.

So, Vauhini, to answer your question of whether or not readers will still read about how it felt to be human throughout it all, I know that I will still be reading. At the very least I will still be feeling human, and I know you will, too. So continue to reach out your hand in your art, and every once in a while, another human will see it, and take it.

_______________________

Contract writing or editing services: https://inklessagency.com/

Watch this blog post as a video essay: