When you ask ChatGPT about the memoir “The Bedwetter” by comedian Sarah Silverman, the artificial intelligence chatbot will provide a detailed summary of every chapter.
Does that imply that it memorized and “read” a stolen copy in its entirety? Or did it scrape so many online discussions and customer reviews about the best-seller or the show it served as inspiration for that it could pass for authority?
After Silverman sued ChatGPT-maker OpenAI for copyright infringement this week, joining a growing list of authors who claim they unintentionally laid the groundwork for Silicon Valley’s raging AI boom, the U.S. courts may now be able to help resolve that issue.
According to Silverman’s claim, the digital copy of her 2010 book that OpenAI used to train its AI models was ingested without her consent and was most likely taken from a “shadow library” of stolen works.
The memoir was copied, as stated, “without consent, without credit, and without compensation.”
It is only one of several incidents that could expose the secrecy of OpenAI and its competitors over the priceless data used to train increasingly popular “generative AI” products that produce original literature, photos, and music.
Additionally, it calls into question the moral and legal foundation of the tools that, according to the McKinsey Global Institute, would boost the world economy by anywhere between $2.6 trillion and $4.4 trillion.
One of the attorneys assisting Silverman and other authors in their pursuit of a class-action lawsuit, Matthew Butterick, claimed that “this is an open, dirty secret of the entire machine learning industry.
They obtain it from these illegal sources since they enjoy book data.
We’re sort of raising the alarm about that entire practice.
Requests for comment regarding the allegations went unanswered by OpenAI.
Similar allegations are made in a different case by Silverman regarding an AI model created by Meta, the parent company of Facebook and Instagram, which declined to comment.
It might be difficult for authors to prevail, particularly in light of Google’s victory in fending off legal objections to its online book library.
In 2016, the U.S. The Supreme Court upheld lower court decisions that denied writers’ arguments that Google’s digitization of millions of books and public display of select passages from them amounted to “copyright infringement on an epic scale.”
According to Deven Desai, associate professor of law and ethics at Georgia Tech, “I think what OpenAI has done with books is awfully close to what Google was allowed to do with its Google Books project and so will be legal.”
The concerns about the tech industry’s AI-building processes have gained traction in the literary and artistic worlds, even though only a small number of people have filed lawsuits, including Silverman and best-selling authors Mona Awad and Paul Tremblay. In a letter sent to the CEOs of OpenAI, Google, Microsoft, Meta, and other AI developers late last month, notable authors Nora Roberts, Margaret Atwood, Louise Erdrich, and Jodi Picoult accused them of using exploitative techniques to create chatbots that “mimic and regurgitate” their language, style, and ideas.