Do you pay $169 for an introductory machine learning e-book with quotes that appear to be constructed?
If not, I recommend giving the purchase Mastering Machine Learning: From basics to advancedpublished in April by Springer Nature.
Based on tips from readers, we checked 18 out of the book's 46 quotes. Two-thirds of them either did not exist or there were quite a few errors. And the three researchers cited in the book confirmed that the work they believed were fake and that the quote contained considerable errors.
“I wrote this paper, but it was not officially published,” said Yehuda Dar, a computer scientist at Ben-Gurion University in Negev. “It's an Arxiv preprint,” the quote states that the paper appeared incorrectly. IEEE Signal Processing Magazine.
Aaron Courville, professor of computer science at the University of Montreal and co-author of the book. Deep learningcorrectly quoted for the text itself, but in the “appears to exist” section he said. “It's certainly not pages 194-201.” And Dimitris Calles of the Helenic Open University in Greece also confirmed that he had not written any cited works listing him as the author.
The researcher who emailed us and asked us to remain anonymous was warned by Google Scholar about the book that cited him. His name appeared in multiple quotes, but none of the works were quoted.
Non-existent, error-prone citations are characteristic of text generated by large language models such as ChatGPT. These models do not search the literature database of published papers such as human authors. Instead, it generates content based on training data and prompts. Therefore, while the quotes generated in LLM may appear legitimate, the content of the quote may be produced.
The book's author, Govindakumar Madhavan, asked for an additional “a week or two” to fully accommodate requests for comment. He did not answer our question asking if he used LLM to generate text for the book. However, he said, “Even human-created text can appear “like AI,” so it remains a challenge to ensure that the content (or problems) are determined whether AI will be generated or not. This challenge is expected to grow only because LLMS is fluent and refined. ”
According to his bio of the book, Madhavan is the founder and CEO of Seaportai and is the author of around 40 video courses and 10 books. The 257-page book contains a section from ChatGpt that states that “the technology raises important ethical questions regarding the use and misuse of AI-generated text.”
Springer Nature provides authors with policies and guidance on the use of AI, Felicitas Behrendt, senior communications manager for the publisher's book, told me via email. “The author recognizes the possibility of using LLM, but emphasizes that submissions must be done with full human surveillance and that use of AI must be declared beyond basic copy editing.”
Master of machine learning No such declarations are included. When asked about the potential use of AI in its work, Behrendt told us: She did not comment on her efforts to ensure that she adheres to AI policies during the Springer Nature editing process.
The LLM-generated citation was at the heart of the controversy over Robert F. Kennedy Jr.'s “Make America Healthy” report and CDC presentation on the vaccine preservative thimeromonkey. In Retraction Watch, our co-founder was quoted in the once-constructed reference in the Australian government's report on the integrity of the research. We've seen fake citations drop research articles, but the list of papers with evidence of private ChatGPT use has grown for a long time and almost certainly represents just a small portion of what we actually do.
On the same day, Behrendt responded to our query, and Springer Nature published a post on his blog entitled “The Integrity of Book Research: Prevention by Balancing Human Surveillance and AI Tools.”
“All book manuscripts are first evaluated by an in-house editor who decides whether to forward or forward the submission for further review,” writes Deidre Hudson Reuss, the company's senior content marketing manager. “Reviewers – subject matter experts – evaluate the quality and originality of the manuscript, ensure its effectiveness and ensure it meets the highest integrity and ethical standards.”
Like a retraction watch? You can make a Tax-deductible contributions to support our workplease follow us x Or blue skiing like us on facebookFollow us on LinkedIn RSS Readeror subscribe to us Daily digestion. If withdrawal is found Not in the databaseyou can Please let me know here. For comments and feedback, please contact us by email [email protected].
process…
success! You are on the list.
Oops! There was an error and the subscription could not be processed. Please reload the page and try again.
