the Second death of the author

2023
Two channel video, sound
20 minutes, 4 seconds

Related works:
Neural Net Dreams

In 1967, the literary critic Roland Barthes famously declared that the author was dead (“The Death of the Author”). For Barthes and other critics like William Wimsatt and Monroe Beardsley, the author’s biography and intentions were irrelevant to textual interpretation, especially because the author might not even know his or her intent or could have changing or multiple intents (“The Intentional Fallacy,” 1946). The reader or the broader culture were the real writers while the author was just a channel, cipher, or collection point. Later criticism has also recognized the idea of text as remix and that authorship can be multiple and collaborative.

Over the last few years, as large language models have grown more sophisticated and accessible, new questions have arisen about authorship in the age of artificial intelligence. When a large language model composes a text, who is the author? The person prompting the model, the architects of the model, the model itself, or the authors of the training texts? How do we understand personal style when a large language model can so easily mimic different voices? What does it mean to co-write with a non-living entity? If Roland Barthes were alive today, what would he have to say about all this?

THE SECOND DEATH OF THE AUTHOR explores these questions and many more in a video essay combining animated photographs of Barthes with a text collaboratively composed by the artist, GPT-3, and ChatGPT with altered excerpts from Barthes’s original essay and Camera Lucida (1980).