Springer Nature is retracting a book on machine learning that had multiple references to works that do not exist, Retraction Watch has learned.
The move comes two weeks after we reported on the book’s fake references.
The link to the information page for the book, Mastering Machine Learning: From Basics to Advanced, now returns “Page not found,” and the text is no longer listed under the book series on computer systems and networks.
Springer Nature confirmed the removal is part of the retraction process: “The book is being retypeset, including with the addition of a retraction note, to reflect its retracted status. Once that has been completed, it will be back on the website (clearly marked as retracted),” Tim Kersjes, Head of Research Integrity, Resolutions, at the company, told Retraction Watch.
As we previously reported, many of the citations at the end of chapters contained significant errors, referencing works or chapters of books that do not exist. We confirmed a few of these errors with four people cited in the references.
These types of errors are common in text generated by large language models like ChatGPT. Using AI without declaration for anything besides copy editing is against Springer Nature’s policies.
As of June 26, the book, which sold for $169, had been accessed 3,782 times, according to its information page.
Following our coverage, Yung En Chee, a quantitative ecologist at the University of Melbourne, contacted her university library, which had the book in its catalog. According to her post on Bluesky and emails we have seen, on July 2, Chee asked the library to “pause access, or ask Springer Nature for an explanation or refund.” The library sent her an update saying the publisher told them they were aware of the problem and were currently investigating what happened. The book was removed from the library’s catalog by July 14.
Kersjes told us Springer Nature is “committed to publishing content that is accurate, trustworthy, and aligned with the highest editorial standards,” and is “reviewing this case carefully so we can understand what went wrong and ensure it does not happen again.”
Springer Nature has retracted books before, as have other publishers.
The author of the book, Govindakumar Madhavan, did not respond to our request for comment. According to his author bio, Madhavan is the founder and CEO of SeaportAI.
Like Retraction Watch? You can make a tax-deductible contribution to support our work, follow us on X or Bluesky, like us on Facebook, follow us on LinkedIn, add us to your RSS reader, or subscribe to our daily digest. If you find a retraction that’s not in our database, you can let us know here. For comments or feedback, email us at [email protected].
And now that Springer Nature has decided to retract this book they will be contacting all purchasers of the book with information on how to obtain a refund from the publisher, right? Right?!?
Yeah, right. That’s about as likely as finding a virgin in a maternity ward. Just ain’t gonna happen.
Your suspicions about the surrogate mothers of the world seem a bit harsh.
Since when is completely erasing any information about book/publication from the publisher’s website considered a retraction? A proper retraction requires a retraction note from the publisher related to the original publication. Otherwise the only way to prove this book was retracted is this post. It’s more like a cover-up.
> The link to the information page for the book, Mastering Machine Learning: From
> Basics to Advanced, now returns “Page not found,” and the text is no longer listed
> under the book series on computer systems and networks.
> Springer Nature confirmed the removal is part of the retraction process: “The book is
> being retypeset, including with the addition of a retraction note, to reflect its retracted
> status. Once that has been completed, it will be back on the website (clearly marked as
> retracted),” Tim Kersjes, Head of Research Integrity, Resolutions, at the company, told
> Retraction Watch.
It’s not like Springer books that should be retracted have the tendency to fully disappear instead. I would not be surprised if the redaction of said retraction note would take quite some time.
I wouldn’t worry about Springer Nature, it is doing very nicely.
https://www.researchinformation.info/news/springer-profits-up-to-e512-million/
“Springer Nature has announced its results for the financial year 2024. Revenue was €1,847 million, representing, on an underlying[1] basis, a 5% increase over the same period in 2023. Adjusted operating profit increased to €512 million, equating to underlying[1] growth of 7%.”
That’s 28% (512€ million ÷1,847€ million), to the nearest percentage, operating profit.
Is it a kind of Retraction?
There’s no notice and explanation for removing this book only from the publisher’s website. It is still available for purchase at Amazon at $322. Where Amazaon brings the book for buyers other than the publisher? Seems a sort of scandal.
Book must have been purchased by many libraries; public, institutions, and universities. The publishers already made expected profit, which is going on through other canals like Amazon, Indigo, etc.
Interesting to see that a book about how to use machine learning was created by machine learning tools and turns out to be essentially useless.
Kind of a meta level commentary on the whole machine learning bubble.
Very true. It is almost poetic. Ai and ML models are useful. However, their blanket acceptance by a large swath of academics and industry folks without much scrutiny aldreay has noteworthy shortcomings. Add this to the list I guess. So the author wrote and used some Ai tools thinking them to be robust and not entirely exploring all the details in his own work. Good example of the pitfalls of using them. Imagine what other glaring issues besides things like citations might be found with further scrutiny.
It seems like ‘retypesetting’ the book is not exactly the solving the problem at hand. Perhaps the publisher should be putting their ample profits to ‘accuracy checking’ instead… might help their credibility.