From time to time, editorial boards quit journals. It happens often enough that I usually don’t pay much attention. But the outgoing editors of the Journal of Human Evolution described a new complaint, involving – surprise! – generative AI.
In fall of 2023, for example, without consulting or informing the editors, Elsevier initiated the use of AI during production, creating article proofs devoid of capitalization of all proper nouns (e.g., formally recognized epochs, site names, countries, cities, genera, etc.) as well italics for genera and species. These AI changes reversed the accepted versions of papers that had already been properly formatted by the handling editors. This was highly embarrassing for the journal and resolution took six months and was achieved only through the persistent efforts of the editors. AI processing continues to be used and regularly reformats submitted manuscripts to change meaning and formatting and require extensive author and editor oversight during proof stage.
This is maddening, because it’s yet another example of gen AI creating problems, never solving them.
I am also baffled. Because usually I can at least understand why a publisher has done certain things. Often the explanation tracks back to, “Cut costs.”
But I cannot for the life of me figure out why a publisher would let a generative AI system loose on a completed, edited manuscript. I cannot believe that is in any way cost-saving.
Generative AI is notoriously expensive. And they were doing this to nominally completed manuscripts, added an additional layer of work. If it were in the hopes of saving costs eventually, you would think there would just be some internal testing, not being unleashed like a rabid raccoon on a working journal.
Nor do I think anyone with any publishing experience would believe it would improve the manuscript.
I am left absolutely confused by what Elsevier is thinking here. But moreover, I am am worried that other publishers are going to try the same thing and we just haven’t heard about it yet.
Other major publishers should see an opening here. They could get out and publicly promise academics that the job of final edit stays with human editors.
Unfortunately, I am having a hard time seeing this happening, as I suspect increasingly the heads of these companies see themselves as data and analytics companies more than publishing companies.
Update, 6 January 2025: Retraction Watch has a response from Elsevier to the resignation. They claim they weren’t using generative AI, but admit they “trialled a production workflow” that caused the errors.
I’m not sure that’s much better.
Obviously, you also hope that workflow changes would be well-thought out enough that they wouldn’t introduce mistakes.
But I’m more unimpressed by Elsevier’s lack of transparency. What were they doing to the workflow that caused the mistakes? That the editors thought this was generative AI suggests that Elsevier did not explain the new workflow to the editors well, if at all.
Update, 10 January 2025: Science magazine is covering this story, and they have quotes from the editors that contradicts Elsevier’s “We weren’t using AI” claim:
According to Taylor and paleoanthropologist Clément Zanolli at the University of Bordeaux, another former editor-in-chief who signed the statement, Elsevier told them in 2023 that AI software had introduced the formatting errors.
Elsevier told Science the same thing they told Retraction Watch: they were testing a new production system. But the statement to Science did not say there was no AI involved, which they said to Retraction Watch.
External links
Evolution journal editors resign en masse to protest Elsevier changes
Evolution journal editors resign en masse
Elsevier denies AI use in response to evolution journal board resignations
Journal editors’ mass resignation marks ‘sad day for paleoanthropology’
