Journal editors resign to protest AI use, high fees, and more

Board members expressed concerns over high fees, editorial independence, and use of AI in editorial processes.

See full article...

 

[imikem]

"The paper has since been retracted, but the incident reinforces a growing concern that AI will make published scientific research less trustworthy, even as it increases productivity."

s/productivity/volume/

FTFY.

 

[internetomancer]

It might, if there are any. If your experience is that it always does, the problem is actually probably that your own reading and writing skills are insufficient to verify the output.

It doesn't always, of course. And instead of just randomly insulting me, you could simply verify my claims. It's not that hard.

One thing that irritates me about AI conversations is that folks talk about them like they exist in another universe, that no decent man would ever investigate.

 

[DarthSlack]

I can't speak for the rest of the government, but when NIH pays for medical studies, the peer-reviewed research must be submitted to PubMed Central where it is publicly available for anyone to access.

https://sharing.nih.gov/public-access-policy

That's a relatively recent development though. I think Collins pushed it through during his tenure as NIH Director. Prior to that, for a lot articles the only thing in PubMed were the abstracts.

It would be great if other funding agencies followed NIH's lead on this.

 

[graylshaped]

I don't think the job should be handed over at all. The human copyeditor should not be fired.

I think you are deliberately misreading my posts for some reason that I can not comprehend.

If you really don't understand me, then just say so, and I can try to elaborate further.

What you propose is:

1) Human does the work.
2) AI checks the work.
3) Human checks the checker to make sure it didn't do something stupid.
4)* Boss notices step 3 is needed only n% of the time and decides Step 3 can be skipped.
5)* Bullshit is put into production n% of the time.
6)* Human gets the blame.
7)* Step 1 is eliminated.

*Optional but highly likely and entirely predictable.

edit: Now, if the process would be to provide to the human access to functional models to aid in the work, then we're all good, as long as we don't tell those telling us how revolutionary "AI" is that we're right back to their status as next-gen spell checkers.

 

[daveishereagain]

This is what happens when producers of a product make profit the most important aspect of producing that product. Product quality and the quality of the work place inevitably suffers. Making money over everything else is running rampant and is pervasive in all phases of culture, business, politics and human relations. The human race is not making progress in these times and continues to encourage our basest natures. Not good, kids.

 

[Writer from Texas]

Maybe the answer is... scientific journals shouldn't be for-profit enterprises. Especially not ones driven by public research.

Amen to that. I long argued for publishing all government-funded research in sources freely available to at least the American public. However, lobbyists for publishers killed that.

 

[internetomancer]

What you propose is:

1) Human does the work.
2) AI checks the work.
3) Human checks the checker to make sure it didn't do something stupid.
4)* Boss notices step 3 is needed only n% of the time and decides Step 3 can be skipped.
5)* Bullshit is put into production n% of the time.
6)* Human gets the blame.
7)* Step 1 is eliminated.

*Optional but highly likely and entirely predictable.

Okay. Yes, I proposed 1-3.

If you think bosses are especially irrational or immoral about this, I suppose that's fair, but it's not something I am suggesting at all.

 

[daveishereagain]

Amen to that. I long argued for publishing all government-funded research in sources freely available to at least the American public. However, lobbyists for publishers killed that.

I have always assumed that, barring classified ones, they were available to the public. It's our money; we should be able to peruse what we paid for.
Learn something new etc,etc.

 

[Clemuns]

This is just a very long way of saying “Graham Hancock was right.”

 

[graylshaped]

Okay. Yes, I proposed 1-3.

If you think bosses are especially irrational or immoral about this, I suppose that's fair, but it's not something I am suggesting at all.

It is actually worse. IRL, the boss wouldn't notice and/or wouldn't care because he or she may actually have a clue. A minion in the beancounter department will notice the n% figure and point it out to his or her boss, who will tell his or her boss, who is also the copy editor's boss, there is an opportunity for cost savings. At the point, the copy editor's boss is in a defensive position, and may have bigger battles to fight.

The theoretical way to avoid this scenario would be to log each and every incident where jabberwocky appears unsolicited, to be armed with a not-made-up catalog of embarrassments avoided. The realistic expectation by this point is that the beancounter minion's boss has already cut the budget of the copy editor's boss to realize the mythical savings.

I offer this not as cycicism, and it is less about anyone being irrational or immoral; it's just observation from my decades of experience seeing how choices over which battles to fight are made on both sides of these types of issues at many different rungs on the ladder. Very few people have the constitution to be Tanner, the Bad News Bear who took on every battle no matter how futile. Most people make decisions on the information they have, and if a boss makes a bad decision, I usually wonder who had his or her ear.

The primary job of the executive officer is not to let the captain run the ship aground--no matter what other responsibilities he or she has. The best boss I ever had (of more than forty, exact number depending on how I count) told his subordinates "Don't let me look stupid." I phrased it differently to my teams: "A major part of your job is to help me avoid silly mistakes."

 

[dmsilev]

I have always assumed that, barring classified ones, they were available to the public. It's our money; we should be able to peruse what we paid for.
Learn something new etc,etc.

Most of them are, actually. Most academics either post PDFs to their lab or group's website or, if nothing else, will email you a copy if asked. University libraries often have central repositories. In many fields, preprints are posted to arxiv.org and similar sites, which differ only slightly from the final published versions.

 

[The Dark]

This is just a very long way of saying “Graham Hancock was right.”

You'd have better luck convincing me that Gene Ray was right.

 

[Aube]

I was slightly misled by the article title into thinking that it was about journal editors at "Evolution" (a major publication in the field of evolutionary biology). Not to say that the subject isn't important, just that to me, the Journal of Human Evolution's not quite spill my (imaginary) coffee without noticing level important.

 

[TVPaulD]

It doesn't always, of course. And instead of just randomly insulting me

I very specifically did not insult you. I observed a potential limitation in your approach that you hadn’t considered and evidence pointing towards it presence. If you felt insulted that is your own issue.

you could simply verify my claims. It's not that hard.

One thing that irritates me about AI conversations is that folks talk about them like they exist in another universe, that no decent man would ever investigate.

Non-sequitur. If you aren’t going to engage with what people are actually saying in response to you, then stop responding in turn. It is extraordinarily rude to simply make up an argument you would prefer to be responding to.

 

[Magazan]

What you and other journalist should be doing with cases like this is to identify the board of directors by name.

It's not ' Elsevier's response was that ...' - it's 'these people in Elsevier decided this'. Make the people responsible stand up and be recognised. Maybe with a bit of journalism we might find that the people on the board of Elsevier have affiliations with certain sponsors, certain funders ...

 

[Uncivil Servant]

That's a relatively recent development though. I think Collins pushed it through during his tenure as NIH Director. Prior to that, for a lot articles the only thing in PubMed were the abstracts.

It would be great if other funding agencies followed NIH's lead on this.

NIH's current policy dates back to at least 2008, per their webpage. Also, this was a statutory requirement resulting from an act of congress, which is appropriate* since congress has sole power of the purse.

There is a more recent 2024 update to their policy, but it does not go into effect until January 1, 2026.


*pun mildly intended


(edited to update the effective date of the policy, whoops)

 

[daveishereagain]

Most of them are, actually. Most academics either post PDFs to their lab or group's website or, if nothing else, will email you a copy if asked. University libraries often have central repositories. In many fields, preprints are posted to arxiv.org and similar sites, which differ only slightly from the final published versions.

OK, thanx for that. How information on taxpayer-funded work isn't available didn't ring true to me.

 

[Nalyd]

Amen to that. I long argued for publishing all government-funded research in sources freely available to at least the American public. However, lobbyists for publishers killed that.

Problem is the open journals charge the PIs directly so it ends up making grants more expensive, and excludes publication by authors with less grant funding or institutional resources. Library subscription funded journals tend to not have page fees or not as high anyway.

 

[orbatos]

Isn't that...the exact job of an editor?


Is there a way this ends that isn't just all the resigned people being replaced by AI and the quality of everything just getting worse?

Ultimately no, they wanted to get rid of the editors in the first place and the investors won't care until profits drop, at which point they will propose raising fees as ads rather than hiring and paying employees.

 

[orbatos]

Problem is the open journals charge the PIs directly so it ends up making grants more expensive, and excludes publication by authors with less grant funding or institutional resources. Library subscription funded journals tend to not have page fees or not as high anyway.

These journals have never cared about this problem and actually cultivate exclusivity on purpose. Remember, they are for profit endeavors, their investors only want stocks to go up and costs to go down.

 

[mgforbes]

Again, I think the answer is to proofread, and then check the result against a good LLM. And I think your time estimates are incorrect. Carefully proofreading a 1,000 words for errors might take 10 minutes. After you read it, you could copy and paste it into an AI, and the AI will find a couple little errors. Time it takes to verify them is less than a minute.

As a former copy editor (and current pedant) I'll point out that "a 1,000" is something you should have caught. "1,000" or "a thousand" would be appropriate, although from a style perspective the latter is preferred. I don't know if your LLM would have noticed that.

Proofreading your own work is devilishly hard. Since you wrote it, you have some expectations of which words were used, and what you meant versus what you said. Spelling errors, particularly those where homonyms are involved, can be hard to spot, and spell checkers are no help there. I have no faith that a LLM would not end up adding errors (of the hard-to-find variety) instead of fixing them.

 

[Nalyd]

Academic librarian here: Elsevier is pure evil, and if you've worked for them for the past ~15 years you don't get sympathy. Working for the Evil Empire that's actively making the world a worse place isn't respect-worthy.

So while it's nice these people resigned, I can't imagine how tf it took so long. I DO know that, when I see "Elsevier" on someone's CV while I'm on hiring committees, that person doesn't get an interview.

While I agree with your sentiment about Elsevier, that’s an awfully absolutist, vindictive, and petty way to treat people. People gotta do things sometimes to get where they are going. And sometimes you do it for a good journal in your field that happens to have been absorbed by the big E, until you can’t anymore.

 

[Nalyd]

These journals have never cared about this problem and actually cultivate exclusivity on purpose. Remember, they are for profit endeavors, their investors only want stocks to go up and costs to go down.

I’m talking about society journals, not for-profits. They charge too for open publishing. A lot.

 

[The Sheep Look Up]

People gotta do things sometimes to get where they are going.

"Look, if you want to be the senior manager at the orphan grinding plant, you're going to have to do your time running the orphan grinder."

 

[MDCCCLV]

Yes-ish. However at scientific journals, editors are there primarily because of their scientific expertise, not their language skills. So it strikes me that a copy editor is probably a good idea otherwise it's hit or miss.



Yes, but Elsevier's been riding that bus for a while now, AI is just slapping rocket boosters on the sides. And, to be honest, if this is the death knell for Elsevier journals, that's probably a good thing for science. Privately owned and published scientific journals are too focused on the money and not the science.

I've had nothing but bad experiences with Elsevier, they're the most difficult to work with of the journals I use.

 

[Mental Gear Reduction]

So while it's nice these people resigned, I can't imagine how tf it took so long. I DO know that, when I see "Elsevier" on someone's CV while I'm on hiring committees, that person doesn't get an interview.

Remember, every one of those people is formerly employed by Elsevier. They might have arrived at the exact same conclusion you did, and bailed out, but you'd still be punishing them for it.

 

[NetMage]

I have no faith that a LLM would not end up adding errors (of the hard-to-find variety) instead of fixing them.

You don’t ask the LLM to re-write the paper so there is no chance it can add an error by itself. It is certainly possible it would suggest corrections that are wrong, but someone would still have to approve them.

 

[Veritas super omens]

"Look, if you want to be the senior manager at the orphan grinding plant, you're going to have to do your time running the orphan grinder."

Well...yeah...those orphan's aren't going to grind themselves, despite the fevered wishes of Elsevier and top tier capitalists everywhere (including the Netherlands).

 

[TVPaulD]

I didn't insult you, I just said that you might be an idiot instead of testing your observation.

Nobody said you were an idiot. Nobody said anything even close to that. I said you perhaps aren’t as good at reading and writing, which are central to copy editing, as you imagine yourself to be. That has the square root of sweet damn all to do with idiocy.

I do not need to test your observation, because as I alluded to already: you did not make a novel assertion nor is your imagined firewall between those you disagree with and the technology in question evidenced anywhere other than your own assertion plucked out of the clear blue sky.

I and others have already mentioned problems directly relating to what LLMs actually do in practice which call your conclusion into question. You just don’t like those observations so you are ignoring them and instead arguing against imagined, weaker comments.

Kindly knock it off.

 

[RichyRoo]

Everyone knows this system is bad, why don't people just start new non-profit journals? Just use GitHub, job done. I really don't understand why these legacy publications are still around

 

[internetomancer]

I do not need to test your observation, because as I alluded to already: you did not make a novel assertion.

I don't know what you mean by this.

You've seen my observation about error rates in major published articles and my suggestion of using AI as a backup checker studied elsewhere?

 

[Wheels Of Confusion]

Everyone knows this system is bad, why don't people just start new non-profit journals? Just use GitHub, job done. I really don't understand why these legacy publications are still around

https://xkcd.com/793/

 

[Mustachioed Copy Cat]

Fuck Elsevier. They continue to make astonishing progress proving that scientific literature copyright is more of a hindrance to scientific progress than a benefit.

 

[ferdnyc]

MOTHERF----.

I am unreasonably distracted by poor copy editing. I can't help it. I see an "it's" instead of "its", or someone claiming that they have "three reasons" we should accept their argument, only to give four, and that's all I can think about -- I can't unsee it.

I've gotten better at recognizing, over time, that this is very much my problem, and not the fault of the authors whose misapplied turns of phrase plague me so. The assumption is that they're doing their best, and after all nobody's perfect, least of all me.

But to hear that a company like Elsevier tells its own editors, "Ahhhh, don't worry about it!" ignites a white-hot fire behind my eyeballs. This is why we can't have nice things.

And the worst part, with these other AI revelations, is that AI can't even rescue us from that minor annoyance. I've seen countless examples of AI-generated content that contains the same homophone confusions, grammatical mistakes, and other writing flubs that you'll find in careless human output. I mean, of course that'd be the case, since that's exactly what the AIs were trained on.

 

[SubWoofer2]

Academic librarian here: Elsevier is pure evil, and if you've worked for them for the past ~15 years you don't get sympathy. Working for the Evil Empire that's actively making the world a worse place isn't respect-worthy.

So while it's nice these people resigned, I can't imagine how tf it took so long. I DO know that, when I see "Elsevier" on someone's CV while I'm on hiring committees, that person doesn't get an interview.

"we find we can no longer work with Elsevier in good conscience"

Thank you Reaperman2 for covering the main question arising from this article. Not "why did they resign?" but "WTF took them so long?"

Elsevier using AI is not causal; it's just the latest symptom in an ongoing disease.

 

[acefsw]

I haven't read all the comments yet, so apologies for repeating anything, but this is what you get with an outfit like Elsevier who have a 40% profit margin and something like 18% to 20% of the "market."

Knowledge should not be a market or profit center, but here we are.

 

[Disordered]

Caeser and Aurich have already answered this question.

It’s about the habituation of users. People are basically lazy, and don’t want to be bothered by having to find a new online community. Considering what Elsevier did in this case, it sure makes one wonder what Conde Nast will do to this website. AI will eventually replace copy editors, but who is it that will judge them? Unlike copy editors, can you fire an AI?

One thing is certain; Conde Nast, like all other MSM, will not offend the new USA Regime. Whether or not that is detrimental to the advance of science and society.