The Problem With AI Chatbots In Literature

An Ultimately Misguided Approach To Reading.

Imagine being able to talk to Shakespeare’s Henry V about leadership in his seminal St. Crispin's Day speech, on the eve of the Battle of Agincourt; or conversing with Leopold Bloom on his views on musical representation in James Joyce’s Ulysses; or asking Elisabeth in Jane Austen’s Pride and Prejudice about what she truly thinks about Mr. Darcy, once and for all.

Recently, TED released a talk featuring Khan Academy founder Sal Khan, who suggested that this is now possible thanks to Khan Academy’s educational chatbot, Khanmigo.

During his presentation, Khan described a situation where a young reader was having trouble understanding why Jay Gatsby is staring at the green light in the distance in F. Scott Fitzgerald’s classic novel The Great Gatsby. After having consulted various sources to no avail, Khan explains, “she realized that she had Khanmigo and that she could talk to Jay Gatsby himself.”

After this follows a transcript of the dialogue between the girl and Khanmigo the chatbot, assuming the role of Jay Gatsby, complete with bizarrely pastiched Fitzgerald-ized language. Here’s an example:

“"Ah, the green light, old sport. It's a symbol of my dreams and desires, you see. It's situated at the end of Daisy Buchanan's dock across the bay from my mansion. I gaze at it longingly as it represents my yearning for the past and my hope to reunite with Daisy, the love of my life.” (06:19)

To me, this approach to using a chatbot is preposterous, regardless of the age and experience of the reader in question. Democratizing education and literature with AI is great. Limiting the human experience in the process is not.

Literature Does Not Need Fixing. Really

From Norton Juster to Saul Bellow, from Aesop to Zadie Smith, literature is ambiguous and messy, leaving room for multiple interpretations and encouraging readers to question, analyze, and actively engage with the text. Here’s a little secret: it is supposed to be that way.

The negotiation of blank spaces in literature are continuous invitations to interpretation, which is a crucial skill in fostering independent and analytical thought; by providing chatbot-generated answers and interpretations, readers are robbed of this essential part of the reading experience and, essentially, human experience.

Literature does not need fixing. The idea of having a chatbot somehow enrich or embellish a literary masterpiece is a provocation and degrading of the writers. When you read books, sometimes you are left with more questions than answers. And that’s a good thing. Really.

The Perils Of Ignoring Literary History

The educational chatbot approach, at least in the example from Khan Academy, completely ignores centuries of knowledge accumulation and research on how novels work.

For example, there’s a strange, anachronistic perspective to all of this which has to do with the intentionality of a literary work, and the separation of author, work, and reader (to simplify grossly). The idea of having an AI chatbot enlighten us on the meaning or purpose of a given text is not only absurd in its own right, it also neglects decades of careful examination of reader roles and functions by Umberto Eco, Roland Barthes, and Roman Jakobson, to mention but a few semioticians.

In reception history, there has been countless instances where political, sexual, and other types of motives in literary works are ascribed directly to their authors, for instance with Vladimir Nabokov’s Lolita or Salman Rushdie’s The Satanic Verses. In this light, it seems peculiarly dated to pursue the notion that a character in a work should be able to contribute to the interpretation of its purpose or meaning, outside of the text in the form of an AI chatbot. Asking a chatbot is comparable to asking an author about the intentions of their work. Academically, this approach died in 1967 with Roland Barthes’ essay The Death of the Author.

Stating the obvious, Jay Gatsby exists only as a character in this particular work of fiction, and he cannot in any way, shape or form contribute to our understanding of the work outside it.

Concluding Thoughts

While the convenience of obtaining immediate and seemingly authoritative answers from AI chatbots may be appealing to inexperienced readers or learners, it is at the expense of exploring literary texts as windows of enlightenment.

Life is too short to experience everything yourself, and reading books is a great way to add significantly to your sum of life experiences, drawing on decades and centuries of successes and failures in all kinds of human endeavours.

The interaction with chatbots in the hope of finding literary meaning risks replacing the processes of grappling with literature - being overwhelmed, forming hypotheses, using one’s imagination, developing critical thinking abilities, amongst other things.

Generative AI has significant potential for learners, and Khan Academy continues to play a key role in this regard.

However, this one was a swing and miss in my…book.