Wiley shuts 19 scholarly journals amid AI paper mill problems
Fake science challenges academic publishing
US publishing house Wiley this week discontinued 19 scientific journals overseen by its Hindawi subsidiary, the center of a long-running scholarly publishing scandal.
In December 2023 Wiley announced it would stop using the Hindawi brand, acquired in 2021, following its decision in May 2023 to shut four of its journals "to mitigate against systematic manipulation of the publishing process."
Hindawi's journals were found to be publishing papers from paper mills – organizations or groups of individuals who try to subvert the academic publishing process for financial gain. Over the past two years, a Wiley spokesperson told The Register, the publisher has retracted more than 11,300 papers from its Hindawi portfolio.
As described in a Wiley-authored white paper published last December, "Tackling publication manipulation at scale: Hindawi’s journey and lessons for academic publishing," paper mills rely on various unethical practices – such as the use of AI in manuscript fabrication and image manipulations, and gaming the peer review process.
The Hindawi affair coincided with the departure of Wiley president and CEO Brian Napack in October, 2023. In its fiscal Q2 2024 earnings report [PDF] last December, Wiley admitted its $18 million decline in research publishing revenue was "mainly due to the Hindawi publishing disruption."
In January, Wiley signed on to United2Act – an industry initiative to combat paper mills.
But the concern over scholarly research integrity isn't confined to Wiley publications. A study published in Nature last July suggests as many as a quarter of clinical trials are problematic or entirely fabricated.
The increasing availability and sophistication of generative AI is not the only factor contributing to the academic publishing crisis, but AI tools make fakery easier.
"The industry recognizes that AI is utilized by paper mills to generate fraudulent content," Wiley's spokesperson told us. "We've recently introduced a new screening technology that helps identify papers with potential misuse of generative AI before the point of publication."
According to a preprint paper released in February, the volume of papers submitted to ArXiv increased considerably in the top three categories between 2019 and 2023 – a period that roughly coincides with the debut of tools like ChatGPT. Computer science papers increased by 200 percent during these four years, followed by physics papers (45 percent) and mathematics (22 percent).
- OpenAI says natively multimodal GPT-4o eats text, visuals, sound – and emits the same
- Prof asks court to protect his Unfollow Everything 2.0 extension from Facebook's ire
- Microsoft teases deepfake AI that's too powerful to release
- H-1B visa fraud alive and well amid efforts to crack down on abuse
Academic publishers, however, appear to want the benefits of AI writing assistance without the downsides. Springer Nature, for example, last October launched Curie – an AI-powered writing assistant intended to help scientists whose first language is not English. Hence calls for better tools [PDF] to detect generative AI output – a call answered by recent efforts to improve AI content watermarking – which some researchers argue won't work.
A Wiley spokesperson characterized the decision to shut the 19 journals as part of its previously announced plan to integrate the Hindawi and Wiley portfolios, and distinct from the paper mill issue.
"As part of this integration, and as is standard practice, we reviewed our journal portfolio and decided to close 19 Hindawi journals that no longer serve their communities," the spokesperson told The Register.
"It is important to draw a distinction between the journal closures occurring now as part of our portfolio integration and the four journals closed in May 2023. The journals closed in May 2023 were heavily impacted by paper mills to such an extent it was in the best interest of the scholarly community to discontinue them immediately."
Meanwhile, in Wiley's fiscal Q3 2024 earnings report, the publisher noted that revenue for its learning division is expected to be toward the higher end of projections due to "Q4 content rights deals for training AI models." ®