In the hearts of readers, ChatGPT has surpassed William Shakespeare. Researchers at the University of Pittsburgh in the United States asked an audience unfamiliar with great literature to compare poems written by ten great authors with those generated by artificial intelligence. And, surprise, most of them prefer to read the latter, reveals a study published in the journal Nature.
The researchers presented participants with poems written by ten authors who marked their era, William Shakespeare, Samuel Butler, Lord Byron, Walt Whitman, Emily Dickinson, TS Eliot, Allen Ginsberg, Sylvia Plath, Geoffrey Chaucer and Dorothea Lasky. At ChatGPT 3.5, scientists asked people to write poems in the style of these authors, which they also had participants read.
When first questioned about the origin of the work, the vast majority of readers were wrong, believing that the poems written by the artificial intelligence were real poems. It is also the latter that they judged to be of better quality, contrary to previous research. Proof that artificial intelligence is now capable of creating illusions.
A simpler and more accessible poem
The authors justify this preference of non-expert readers with its “accessibility”. Poetry generated by ChatGPT “communicates emotions, ideas, and themes more directly and in easy-to-understand language,” they say. Readers also expected less well-written texts and tended to think that a poem they liked was necessarily written by a person.
However, the “complexity and opacity” of poetry written by human beings ”are part of the charm of poems”, regret the authors who ask governments for better regulation and transparency of the uses of artificial intelligence. An all the greater need given the difficulties readers have in identifying poems written by ChatGPT.
“A poem is more than an algorithm,” underlines the poet Joelle Taylor, author of C+nto & Othered Poems, winner of the TS Eliot Prize with the Caretaker. It is meaning, empathy, revelation, reversal, dissent, passion and surprise: poetry is what happens in the space between logic and chaos. »
What are the implications of AI-generated poetry on traditional literary values?
Interview Between Time.news Editor and Literature Expert
Editor: Welcome to Time.news! Today, we have the pleasure of speaking with Dr. Emily Reid, a literary scholar from the University of Cambridge, who’s been closely following trends in AI-generated literature. Dr. Reid, thank you for joining us.
Dr. Reid: Thank you for having me! It’s a fascinating topic, and I’m excited to dive into it.
Editor: Absolutely! Recently, a study from the University of Pittsburgh revealed that readers prefer poems generated by ChatGPT over those written by literary giants such as Shakespeare and Emily Dickinson. What do you make of this finding?
Dr. Reid: It’s certainly surprising at first glance, right? However, it underscores a significant shift in how we perceive literature today. The study highlights the vast improvements in AI’s ability to mimic styles and produce coherent and engaging texts, which can sometimes resonate with contemporary readers more effectively than traditional poets.
Editor: So you think it’s about resonance? Could it be that modern readers find AI-generated poetry more relatable?
Dr. Reid: Exactly! Readers often seek connection and relevance in poetry. AI, when programmed with large datasets of diverse literary styles, can generate works that may reflect modern sensibilities or use language that feels more accessible. The findings suggest that aesthetic preferences might evolve based on exposure to new forms of creation.
Editor: That’s a compelling point. The study especially noted that readers often mistook AI-generated poems for those from renowned authors. What does this say about our understanding of authorship and authenticity in poetry?
Dr. Reid: It raises profound questions about authorship and authority in literature. Traditionally, we have held great authors in high esteem, connecting their identities deeply with their work. However, if readers struggle to distinguish between AI-generated and human-created poetry, it suggests that the emotional or intellectual responses to literature may transcend the identity of the creator.
Editor: Fascinating! In your opinion, does this mean that AI has a place in the future of literature, or does it pose a threat to human writers?
Dr. Reid: I believe it can coexist. AI can serve as a tool for creativity rather than a replacement. For instance, human poets can use AI to experiment with styles or overcome writer’s block. It can inspire new works and rethink what poetry can be, while still allowing for the unique human experiences and emotions that define great literature.
Editor: That’s a refreshing perspective. How should educational institutions and literary communities adapt to this change?
Dr. Reid: It’s crucial for educators to incorporate discussions about AI in literature classes. We should explore both the capabilities and limitations of AI in writing. This way, students can develop critical thinking regarding the sources of their inspiration, the nature of creativity, and the distinctions between human and machine-generated work.
Editor: Agreed! Before we wrap up, what do you think the future holds for readers and writers as we continue to navigate this intersection of technology and art?
Dr. Reid: The future is undoubtedly intriguing. As technology advances, I see a world where AI could challenge our perceptions of creativity, pushing writers to explore new themes and forms. It could democratize poetic composition, allowing those who may not traditionally consider themselves poets to express their ideas through AI. The dialogue between humanity and technology in literature will be enriching and transformative.
Editor: Thank you, Dr. Reid, for sharing your insights with us. This conversation has certainly opened up many avenues for thought regarding literature’s future in the age of AI.
Dr. Reid: Thank you for having me. It’s always a pleasure to explore such a vital and evolving topic!