Grace Lapointe’s fiction has been published in Kaleidoscope, Deaf Poets Society, Mobius: The Journal of Social Change, and is forthcoming in Corporeal Lit Mag. Her essays and poetry have been published in Wordgathering. Her stories and essays—including ones that she wrote as a college student—have been taught in college courses and cited in books and dissertations. More of her work is at https://gracelapointe.wordpress.com, Medium, and Ao3.
In the NYT, Emily Eakin analyzed “the plagiarism plot,” using recent, popular examples like R. F. Kuang’s 2023 novel Yellowface. As Eakin mentioned, Yellowface is also about cultural appropriation and white privilege. The protagonist, June, steals deceased novelist Athena Liu’s last manuscript. June and her editors also add racist stereotypes that weren’t present in Athena’s original text. Even characters who are unaware of her plagiarism think this is not June’s story to tell. Originality and ethics are timeless themes in art, and plagiarism is just one aspect of those topics. It’s worth considering why privileged authors get away with plagiarizing marginalized authors in real life, who face more barriers to publication.
Since ChatGPT launched in late 2022, it has changed the way people think about plagiarism. As students’ awareness of ChatGPT increases, so does their use of it in schoolwork. However, it’s not yet clear whether there’s been a noticeable increase in direct plagiarism: copying or failing to credit individuals’ work. This may mean that AI-generated content (all of which is stolen) is making theft less detectable. OpenAI admits it trains ChatGPT on copyrighted work, so, I believe there’s no ethical way to use it.
I take plagiarism personally. My work is embodied and comes from my lived experience as a disabled person. I frequently write about my personal experiences of ableism, sexism, and sexual harassment, and they inform my analysis in general. Long before ChatGPT, web crawlers copied my work and stole huge sections of it, with made-up bylines and no credit or link. Individuals have also used my exact words and ideas without quoting or citing them. After I reached out privately, everyone involved edited the documents to cite me properly. This incident made me suspect that the software used to detect plagiarism in the document was unreliable. It scanned academic databases, but not blog posts, which may be why it didn’t flag a long passage improperly copied from my work.
It’s easy for people to plagiarize unintentionally by citing sources incorrectly or forgetting where they read something, and I appreciate it when they fix mistakes quickly. People often coincidentally have similar experiences, thoughts, or phrasing, unlike a computer program, which has no thoughts of its own. I give generative AI no such leeway because it’s built on stolen material.
Maybe a backlash to generative AI is beginning. According to a 2023 US court ruling, AI-generated art can’t be copyrighted. This is a great start, but the specification that art with no human input can’t be copyrighted might be too narrow to cover many uses of AI programs. Authors and music publishers whose work was used to train generative AI without their permission have filed class action lawsuits. I hope that all artists get compensated fairly for their work being used without their permission.
I think it’s still too soon to say whether ChatGPT has contributed to increased plagiarism in traditionally published books. Many systemic issues in the publishing industry predate the invention of programs like ChatGPT. For example, authors are legally responsible for the originality of their own books. Nonfiction authors are also personally responsible for their books’ accuracy. Authors often hire their own fact-checkers, research assistants, and sensitivity readers. Book lovers who don’t work in publishing often don’t know this and assume publishers, not only authors and freelancers, are responsible for books’ originality.
Alice Nuttall wrote on BR in 2022 that it often takes unusually observant readers to notice plagiarism. She also wrote that checking books for plagiarism against a database of published work would risk piracy. Years later, lawsuits alleged that OpenAI used a database of pirated, copyrighted materials to train its language learning models. Ironically, technology that hypothetically could detect plagiarism instead may have been used to produce and profit from copyright infringement on a large scale. Similarly, recent attempts to label human-made books could backfire someday, if LLMs use these books for more training.
English (US) · 