AI & The Death of the Author

AI and Authorship

The meaning of writing doesn't depend on who the author is, even if it's AI. David J. Gunkel, a professor at Northern Illinois University and Łazarski University in Warsaw, wrote that people want to know who wrote something to judge its truthfulness. If the bio says the author is a communication studies professor, one might trust what they say about large language models. But with AI like ChatGPT, Claude, or DeepSeek writing, it's hard to tell who the author is. An algorithm writes the text, but a human prompts it. So, is the author the algorithm, the human, or both? Why does it matter?

The End of the Human Writer?

Since ChatGPT's launch in 2022, many have worried about the end of human writers, fearing LLMs will take over writing or that humans will rely too much on them. One journalist worried that this technology would render people “wordless, thoughtless, self-less” and emphasized the tremendous stakes, dwarfing any previous wave of automation.

However, some believe LLMs signal the end of the author, which isn't necessarily bad. These machines can free writers and readers from the author's control.

The Modern Concept of the Author

Most people think of an author as someone who writes a text and is responsible for what it says, like William Shakespeare, Jean-Jacques Rousseau, or Virginia Woolf. However, this idea is modern. Roland Barthes, a French literary critic, wrote in his 1967 essay “The Death of The Author” that the concept of the author emerged in Europe around the mid-16th century. Before that, people wrote texts, but assigning responsibility to a single person wasn't common. Many great works of literature circulated without an author.

The modern period saw intellectual and cultural changes that Michel Foucault called a “privileged moment of individualizationin the history of ideas.” The Protestant Reformation created individualized faith, and René Descartes' philosophy centered on self-conscious thought. The concept of personal property as an individual right also emerged. Barthes and Foucault argue that the concept of the author comes from these developments.

Copyright Law

The author also became important legally. In 18th-century England and its North American colonies, copyright law made the author responsible for their work. This wasn't about artistic integrity but about the printing press, which allowed texts to be freely distributed. Sven Birkerts explains in “The Gutenberg Elegies” that individual authorship became important when print replaced orality. When copies of text became easily accessible and profitable, identifying the author became crucial for commercial reasons.

Literary Authority

The idea of the author had significant effects on literary theory. Barthes wrote that “When the Author has been found, the text is ‘explained.’” Authority shifted from the writing itself to the author's thoughts and intentions. Readers then tried to understand the author's original intent. Critics and philosophers agreed with Descartes that reading good books was like “having a conversation with the most distinguished men of past ages.” The author, like other authority figures, was often a white man.

Aristotle saw writing as a symbol of mental experiences. Claude Shannon and Warren Weaver's communication model also showed a unidirectional flow of information from author to reader. The best writing was seen as clear and direct, allowing readers to understand the author's thoughts.

The Death of the Author

Barthes argued that the author's role could end. The “death of the author” means the end of the author as the sole authority. LLMs produce text without a living voice, making the text unauthorized. The U.S. Court of Appeals denied authorship to AI, emphasizing this point.

Critics call tools like ChatGPT “stochastic parrots” for mimicking speech without understanding. The disruption of authorship and authority has disturbed many. However, the concept of the author is a social construct, not a natural phenomenon. After the “death of the author,” meaning comes from the reader's experience. Readers “fabricate” what they think the author wanted to say. When reading “Hamlet,” we interpret it and project our interpretations onto Shakespeare. The authority shifts from the author to the reader.

Barthes wrote that a text's unity is in its destination, the reader, not its origin. The death of the author is the birth of the critical reader. This shift explains how LLM-generated content has meaning. While LLMs don't truly comprehend meaning, their writings can be meaningful through interpretation. This applies to all writing, including this essay.

The Concept of Meaning

LLM AI questions the concept of meaning. Words are assumed to refer to real things because an author uses them to say something about those things, as Aristotle suggested. LLMs lack this ability, manipulating words without knowing what they refer to. However, structural linguistics challenges this view, arguing that meaning comes from the relationship between words. In a dictionary, words are defined by other words. Jacques Derrida stated, “Il n’y a pas de hors-texte,” meaning “There is nothing outside the text.” This is especially true for LLMs, where meaning is derived from the texts they are trained on.

Therefore, the criticism that LLMs only circulate words without real-world reference might be inaccurate. LLMs are structuralist machines that disrupt classical semiotics. Generative AI has a large impact on the world. Linguists, philosophers, and AI experts often reassert undermined concepts of authorship and authority. These traditional ways of thinking exert their influence as if they were normal and beyond question.

Artificial Intelligence

Our misunderstanding comes from how we see “artificial intelligence.” AI outputs are seen as signs of intelligence or the lack thereof. LLMs produce text without intelligence, destabilizing the rules. This reveals the philosophical significance of LLMs: they write without speaking, and their truth isn't assured by an intention to say something.

This challenges our understanding of writing and truth. However, it also allows us to think beyond Western metaphysics. LLMs don't threaten writing or truth but a limited understanding of them. They reveal the limits of the author and open opportunities to think differently. Readers can question the source and authority of any text, even one claiming to be human-generated. The difficulty of knowing the author is a condition of all writing, exposing authorship as a fiction.