New Certification for Human-Written Books
As scandals surrounding the use of artificial intelligence by professional authors continue to proliferate, a British startup has developed a method for publishers to certify that their books are “organic,” meaning they are authored solely by human beings without any machine intervention.
On Wednesday, Books by People announced that it has received approval from the UK Intellectual Property Office to begin distributing a seal that indicates a published book has been written by a human. This seal, designed to resemble a fingerprint, will be displayed on the book’s cover.
Esme Dennys, co-founder of Books by People, commented, “If people choose to write books with AI or seek them out for their enjoyment, that is their prerogative. However, if you want to read something penned by a real person, you should easily be able to find it. We are simply trying to provide that option.”
Ensuring Authenticity in Publishing
Previous labelling systems, such as “Human Authored,” relied on the honesty of authors, who pledged not to use AI in their writing process. To enhance the credibility of its certification, Books by People requires an authorship declaration, scrutinises each manuscript using proprietary software, and manually reviews the initial drafts and research materials provided by the author and publisher for “indicators of humanity.”
Dennys explained, “We analyse all elements that allow us to determine with greater certainty whether a human has written the work, meaning it has been created through a natural process of drafting and revision.”
Books by People will charge publishers a fixed membership fee, along with an additional cost for each book submitted for certification. Thus far, five publishers have signed up, and the startup anticipates that the new seal will appear on its first book by the end of the year.
Impact of AI Scandals on Trust in Literature
This year has seen the publishing industry embroiled in a series of scandals concerning the alleged use of AI by authors. A literary agent cancelled a $2 million publishing contract, a publisher withdrew a novel from shelves, and readers claimed that an award-winning short story had been generated by AI.
Industry insiders have indicated that writers and publishers are not necessarily concerned about being replaced by AI, believing that, in the long run, few readers will be interested in books produced by machines.
Author Ted Chiang likened the situation to standing at the end of a conveyor belt, watching items fall off the edge. “Once you know that, it inhibits you from truly engaging emotionally with whatever you’re handling from that conveyor,” he stated. “The knowledge of how something was created cannot be separated from your experience.”
Maintaining Integrity in the Publishing Process
However, the repeated scandals may undermine public trust in the unspoken rule of the publishing world: that every book on the shelves has been written by a real person who has not committed plagiarism. Sam Jordison, co-director of independent publisher Galley Beggar Press, noted, “The publishing world has always relied on trust between publisher, writer, and reading public. Instances where that trust has been betrayed have been rare.”
He added that the notion of reading AI-generated text that pretends to be human is almost repugnant, feeling deceptive in nature. The concept of certification for human-written books has garnered support from various authors and publishers. Markus Dohle, former CEO of Penguin Random House, praised Books by People, referring to human storytelling as “the soul of our industry.”
Potential Risks of AI Detection Systems
Despite the enthusiasm surrounding the initiative, the widespread implementation of AI detection could backfire on authors, as the risk of false positives could damage their reputations. Tom West, CEO of Publishers’ Licensing Services, cautioned against the potential for a witch hunt. “Clearly, the issue lies in the details,” he said.
Pangram, a New York-based startup leading the way in AI detection, claims it can identify AI-generated texts with a 99.7% accuracy rate. However, authors have raised concerns about the reliability of this method, which also employs AI. In September, debut author Jerry Falade alleged that the detection software misinterpreted his Nigerian dialect as evidence of AI use, leading it to classify his book as 97% AI-generated. “It takes months, even years, to write a book,” Falade wrote in a recent essay. “AI detection software can ruin a lifetime of work in just two minutes.”
Reassessing the Norms of Authorship
Others have warned against the normalisation of excessive scrutiny of authors. They argue that labelling systems should address the responsibility of platforms and publishers to disclose when content is AI-generated, rather than questioning when it is human-written. “Authors are constantly interrogated about the origins of their ideas and work. I think we should oppose that,” West suggested. “It should be assumed that a book has been written by a human.”
The verification process established by Books by People, alongside the seal approved by the Intellectual Property Office, differs from Pangram’s technology-driven approach. Books by People emphasises document analysis and communication records to look for signs of humanity. Dennys remarked, “I wouldn’t call it AI detection. We realised quite early on that as AI models improve, entering that race will never be sustainable.”
Moving Towards a Trustworthy Publishing Future
When Books by People is uncertain whether evidence indicates the use of AI, it opts to favour the author. “In public processes subjected to rigorous scrutiny, the only thing we can do is ensure that a book has undergone that process,” Dennys stated. “The core principle of our organisation is to prevent [writers] from being unfairly accused.”
Some authors believe it is essential for publishers to highlight their commitment to human-written literature as a non-negotiable standard for the industry, regardless of the imperfections in compliance methods. “It is as important for publishers to have a ‘no cheating’ policy as it is for universities, even if there is no perfect method to enforce it,” Chiang argued. “It is a statement of values.”