A London librarian has analyzed millions of articles in search of uncommon terms abused by artificial intelligence programs
Librarian Andrew Gray has made a “very surprising” discovery. He analyzed five million scientific studies published last year and detected a sudden rise in the use of certain words, such as meticulously (up 137%), intricate (117%), commendable (83%) and meticulous (59%). The librarian from the University College London can only find one explanation for this rise: tens of thousands of researchers are using ChatGPT — or other similar Large Language Model tools with artificial intelligence — to write their studies or at least “polish” them.
There are blatant examples. A team of Chinese scientists published a study on lithium batteries on February 17. The work — published in a specialized magazine from the Elsevier publishing house — begins like this: “Certainly, here is a possible introduction for your topic: Lithium-metal batteries are promising candidates for….” The authors apparently asked ChatGPT for an introduction and accidentally copied it as is. A separate article in a different Elsevier journal, published by Israeli researchers on March 8, includes the text: In summary, the management of bilateral iatrogenic I’m very sorry, but I don’t have access to real-time information or patient-specific data, as I am an AI language model.” And, a couple of months ago, three Chinese scientists published a crazy drawing of a rat with a kind of giant penis, an image generated with artificial intelligence for a study on sperm precursor cells.
In general, if it passed peer review it shouldn’t matter how it was written.
The fact the blatant examples apparently made it past peer review show how shoddy the process is though.
The editing too. I worked as an editor for academic journals and newspapers about 25 years ago, and nothing like these “blatant” examples would get anywhere near print. We’d remove clichéd language too. Everyone seems to have stopped proof reading and editing.
It’s because all the management level types above the editors all got the brainwave to fire the editors and “just use AI” instead, and entirely failed to understand that the technology is in its infancy and really cannot be considered reliable for things like this, especially if it’s used in such simplistic plug-and-play fashion.
Publishers not proofreading was long before AI came into play. I’ve noticed it for at least a decade now.
There was a whole season of The Wire that was dedicated to the theme of news publications demanding that more be done with less as budgets were cut. Craigslist was a major factor in the trend as it cut revenue severely for local publications.
It started before the rise of AI though imo. So I think it was just an easier out
Good thing the cost to publish went down /s
The academic paper system has been in trouble for decades. But man, the last 10-20 years seems to have reached such an abysmal state that even the general public is hearing about it more and more with news like this, along with the university scandals last year.
I hate how much time and energy is wasted on this bullshit…
You’d think the smartest people around would come up with a be better system than this. I mean they did, but some of the highest decision-makers have big incentives to keep things as they are. So mark that one more on the “capitalism ruins everything it touches” scoreboard.
¯\(ツ)/¯
Incentives matter in any system. The incentives are perverse right now.
So, not very meticulous?
I don’t find these journals’ processes commendable.
deleted by creator
It’s not the reviewer’s fault! When they asked ChatGPT to peer review the paper it found nothing wrong.
Things are very people specific as to what gets through some Journals are known to have lax review but high publication costs. These “predatory” journals and other nepotism stuff has been an issue for a while. The scientific community wants to tackle these issues but it’s been hard to make any real progress. Covid politics and now AI have really not helped.