Working at the Centre for Writing at NYU Abu Dhabi is extremely fun and rewarding. Helping with writing from across different genres, levels, and majors builds an appreciation for the diversity of writing style and text — sure. But the best part is when people do not have any writing to review, just raw thoughts, and you help them put their unstructured, jumbled thoughts into beautiful, clear, and cogent prose. But I am not lucky enough to have appointments all the time. When I am not booked, David Allway, our supervisor, an avid podcast listener and the funniest person on campus (pay raise?), asks me to just exist. Obviously, it would be philosophical blasphemy to just exist and not think. After all, I exist, therefore I think. So, this unstructured column will be about some recent ruminations on the philosophy of writing, done at the Center for Writing NYUAD, as a peer tutor just existing, and therefore thinking.

We often think that writing is about expressing how one feels. Experience is often thought to be at the center of any writing — at least writing that is remotely meaningful. For example, as a reader, I find myself projecting how the author might have felt as they wrote a particular sentence. Most of the time, it is not such a meta-analysis of what the writer might have been feeling. Often, and usually in stories, it is more about how the character is feeling. People often argue that a good author becomes their character as they lead their characters through a story. An author strips their identity, takes the role of the character, and tries to act in the world as if they were the character. It is not to say that the author’s actual identity does not bleed into the characters. Characters are heavily influenced by authors; there is no disagreement there. But the effort an author puts into the character disembodies them, helping them transcend the mere reality. When the author creates a writing, the author makes some part of it tangible in what they write: a thought, an event, a feeling, a bother. As a reader, you try to simulate through the piece of text what the author wished to convey.

It seems then that the author is mostly what we perceive. The writer could have (and most likely) had a completely different feeling when they wrote the piece than what I project it to be today, while I read the piece. And in some ways, we are okay with that. We appreciate different interpretations of a single text. Like poetry, some stories are appreciated for multiple meanings, based on the reader. Unlike academic writing, ambiguity, at least of some kind, is taken to be a good thing in the case of creative writing. If the author is just our projection, then what is wrong with writing produced by Large Language Models? As long as we can ascribe the feeling to the author (non-existent when it is produced by an LLM), should we not have a similar experience as reading a text written by a writer? Or if it is a problem of knowledge, i.e., if you think knowing that a text is written by an LLM makes you incapable of imagining how the characters, and therefore the author, might have felt, is it at least conceivable that LLM writing can be authentic writing until you do not know the source of the writing? Perhaps, the knowledge of the source collapses the capability of projecting the phenomenology / the experience associated with the writing. But in cases where one is devoid of that knowledge, it seems conceivable that LLM writing might not just pretend, but be actual writing, precisely because the reader can project the experience of writing into this abstract author they imagine exists. This is all well and deeply romantic. However, some views wish to segregate and make opaque, the writer from the writing.

Roland Barthes, a French literary theorist, wants the author dead — at least metaphorically. “Writing is that neuter, that composite, that obliquity into which our subject flees, the black-and-white where all identity is lost, beginning with the very identity of the body that writes.”, writes Ronald Barthes, in The Death of the Author. For Barthes, an “author” is a much more modern character than we would like to think; a product of capitalist ideology, linked intricately to private property and ownership. Barthes thinks that a reader is the one who has been marginalized for ages, when the unity of a text lies not in the source (the author) but in the destination (the reader).

“To assign an Author to a text is to impose a brake on it, to furnish it with a final signified, to close writing.”, says Barthes. The true site of writing is then not the author, but the reader. However, unlike the author, the reader does not force the writing to have a single “secret meaning” to be deciphered; the reader does not claim to be personal. A reader is one without history, without biography, without psychology, says Barthes— a reader is the space where multiple interpretations of the writing intersect. Therefore, Barthes makes one of his most famous claims: “The birth of the reader must be requited by the death of the Author.”

Now that writing can be done by the LLM, where the author is not clearly defined, is it a good thing for Barthes? This erasure of the author due to AI, is it the birth of the reader, and the death of the Author that Barthes wanted?

On a closer inspection, it seems not. What Barthes wanted was just the removal of excessive importance on the personality of the author in the writing, not the collapse of writing itself. Barthes wanted to suppress the author for the sake of writing and to restore the status of the reader. Barthes did not want an author, but he did want a scriptor, a performer, one who would synthesize the existing ideas in the world and bring them forth to the reader. What LLM is doing is exactly the opposite. By commodifying text, diminishing each sentence to the levels of small tokens to optimize for the next best prediction, it is not just erasing the author; Barthes would probably think an LLM erases meaningful writing, too.

How does this relate to the earlier idea that an LLM writing could potentially be “actual writing” as long as an abstract notion of an author can be used to project the feelings the characters might have? It seems to me that as AI writing becomes increasingly more prevalent, we will, as a culture, start having skepticism about whether or not any piece is human-generated. The skepticism then is enough to collapse any writing into “non-writing”, even in cases where the writing could actually have been human-written. That state of the world where every form of writing by default is questioned on whether it is actual writing sounds deeply troubling to me.

Then perhaps Barthes was wrong, or the advent of AI has made Barthes less relevant. What we need right now, it seems, is the reawakening of an author. We need to focus not on writing but on the agent of the writing, the producer, the experiencer, one who we can imagine to be feeling and wanting to express. In the age of AI, the death of the author is followed immediately by the death of writing.