The number of books written by ChatGPT’s AI chatbot is likely much higher, as Amazon’s policy does not explicitly require authors to disclose information about the use of artificial intelligence.
Reuters writes that the authorship of the chatbot is indicated in 200 books in the Kindle Store. This is one of the latest examples of use of ChatGPT, which gained public attention after OpenAI made it freely available in November.
“I’ve seen people make a whole career out of it. The idea of writing a book finally became more possible,” said Brett Shickler, a Rochester, New York-based seller who has published a children’s book on the Kindle Store.
Schickler’s creation, Wise Little Squirrel: A Savings and Investing Story, is a 30-page children’s book written and illustrated by AI that retails for $2.99 for a digital copy and $9.99 for a print version. Although Brett says the book has made him less than $100 since it was published in January, he spent several hours creating it. The author used ChatGPT prompts like: “write a story about a dad teaching his son financial literacy”
Other examples of AI-generated Kindle Store content include the children’s story The Power of Homework, the poetry collection Moon of the Universe, and the interstellar brothel sci-fi epic Galactic Pimp: Volume 1
This is something to worry about. These books will flood the market and many writers will be out of work. Authors and platforms must disclose information about how such works are written. Otherwise, we’re going to end up with a lot of low-quality books,” said Mary Rosenberger, executive director of the Authors Guild.
Meanwhile, American online fantasy and science fiction magazine Clarkesworld Magazine has suspended submission of short stories after a flood of articles suspected of using AI without warning. The paper’s editor, Neil Clark, says he identified the chatbot-generated stories by “some very obvious patterns,” but did not specify which ones.
“I can say that this month the number of spam messages leading to blocking has reached 38%. Although it was simple to reject and ban the submission, the number of such submissions is increasing, so other additional measures are required. More dangerously, the technology will only get better, so it will be more difficult to identify the works of AI,” he says.
Clarkesworld currently prohibits submissions of stories “written by or co-authored by AI”. This month, the publication banned more than 500 users for sending content allegedly created by artificial intelligence. Clarkesworld pays 12 cents per word, which is an obvious quick cash opportunity.
From the fact that I can tel, it is not about consent. This is about the opportunity of making a quick buck. That they all can be.
— clarkesworld (@clarkesworld) February 20, 2023
At the same time, questions arise again regarding the veracity of information provided by artificial intelligence, or possible plagiarism. For example, ChatGPT and Bard can “hallucinate”, that is, confidently pass a lie as a fact. In addition, chatbots learn from human-generated content – often without the knowledge or permission of the original author – and sometimes use identical syntax to the source material.
Last year, tech publication CNET used its own artificial intelligence model to generate at least 70 economic articles. In addition to the clever approach to reporting AI authorship (only if the reader clicked on the author line), the works contained numerous factual errors and almost identical phrases from other sites. As a result, CNET issued a patch and suspended the tool, but one of its sister sites has at least experimented with using it again.
Source: Engadget