🐭🍆
February 16, 2024 1:24 PM   Subscribe

The rat with the big balls and the enormous penis – how Frontiers published a paper with botched AI-generated images "These figures are clearly not scientifically correct, but if such botched illustrations can pass peer review so easily, more realistic-looking AI-generated figures have likely already infiltrated the scientific literature." Contains a mildly NSFW image. posted by What is E. T. short for? (59 comments total) 16 users marked this as a favorite


 
dck
posted by Horace Rumpole at 1:26 PM on February 16 [4 favorites]


Retat
posted by capricorn at 1:28 PM on February 16 [1 favorite]


Dissilced
posted by slater at 1:30 PM on February 16 [2 favorites]


Weirdest thing is the paper doesn’t appear to be fake? And the authors site the source of the images as the AI generator.

So they… needed some imagery to jazz this thing up and trusted the generator? Maybe? Are these things just spot illos to set the mood?

And nobody thought this was a bad idea.

Weird.
posted by Artw at 1:32 PM on February 16 [3 favorites]


Ahhhhh, frontiers, mdpi and hindawi , the trifecta of trash publishers polluting the scientific literature with 'peer reviewed' 'journals'.

A colleague asked me to describe a recent paper in one of these paragons.

I described it thusly: "soft, strong and very adsorbent".


*hindawi name now retired by wiley, their owners. watch out, wiley readers!
posted by lalochezia at 1:32 PM on February 16 [29 favorites]


I always eat my sterrn cells with a spoon
posted by shesdeadimalive at 1:33 PM on February 16 [6 favorites]


FARK.COM
posted by credulous at 1:36 PM on February 16 [2 favorites]


@Artw, there is a comment at the end of the blog that notes that the text also seems generated, and gives some good examples.
posted by sonofsnark at 1:40 PM on February 16 [5 favorites]


Aw well fair enough, all bullshit all the way down. But they do cite the images as being bullshit.
posted by Artw at 1:44 PM on February 16 [1 favorite]


"Down in a hole and they've put all the stones in their place
I've eaten the sun so my tongue
Has been burned by the codpiece"
posted by clavdivs at 1:45 PM on February 16


Get that rat jak'd, stat.
posted by grumpybear69 at 1:45 PM on February 16 [2 favorites]


To be honest, my first reaction on seeing this was that it was an inventive interpretation of rat anatomy. It's from a paper about semen generation, after all, so it's unsurprising that it prominently features the rat's genitalia. Blending that aspect of the illustration with the rat itself just seems like artistic license.
posted by SPrintF at 1:47 PM on February 16 [4 favorites]


What a dick.
posted by brundlefly at 1:47 PM on February 16 [2 favorites]


Frontiers: How We Publish
Rigorous peer review

Editorial decisions at Frontiers are guided by clear acceptance and rejection criteria during a single anonymized, rigorous peer review. Editorial independence of acceptance decisions is assured by the involvement of external experts as editors and reviewers. Our in-house publishing professionals quality-control and monitor the process, may reject manuscripts at any point during the review and validate final editorial decisions to ensure our quality standards and policies are being adhered to.
posted by Nelson at 1:49 PM on February 16 [2 favorites]


"Our in-house publishing professionals are also turned on by huge rat dong; in retrospect, this was inevitable."
posted by phooky at 1:51 PM on February 16 [14 favorites]


ratfaked again
posted by snuffleupagus at 2:02 PM on February 16 [6 favorites]


Cf.
posted by Halloween Jack at 2:02 PM on February 16 [1 favorite]


I know this a terrible thing...

But I just can't stop laughing. This is my kind of AI
posted by Windopaene at 2:08 PM on February 16 [8 favorites]


OK I shrieked with laughter when I saw the actual figure. I'm glad my office door is closed.
posted by Tesseractive at 2:16 PM on February 16 [5 favorites]


It’s very annoying to have to think of a scientific paper as NSFW. I really wish the popular coverage didn’t lead with the genitals because that’s the least of the trouble here.
posted by eirias at 2:34 PM on February 16 [5 favorites]


I give FIGURE 2 ("Diagram of the JAK-STAT signaling pathway" or "How to make sprinkled donuts") an 8.4 on the KAWAII scale. But seriously, this whole thing is such an insult to people who actually work hard and try their best.
posted by jabah at 2:39 PM on February 16


ben, the two of us need look no more
we both found what we were looking for ...
posted by pyramid termite at 2:50 PM on February 16 [7 favorites]


Pursuant to lalochezia above, this is your casual-yet-connected-to-this-story reminder that Wiley paid $298 million to buy Hindawi in 2021.

(lol)
posted by aramaic at 2:58 PM on February 16 [6 favorites]


Metafilter: lead with the genitals
posted by credulous at 3:04 PM on February 16 [8 favorites]


it's unsurprising that it prominently features the rat's genitalia

Me too. I haven't read the paper as I am, much to my regret, completely unqualified to pontificate on mousecock, but looking at the illustration I can't help but think "seems like a not insane attempt to represent something small relative to the attached animal when you want focus on that thing and show a lot of detail". I mean, it's not a *good* visual (though it's infinitely better than the 'just do some line drawings in pic or LaTeX and call it a day' in the papers I usually read), but I definitely know what the authors think is important. And sure, it's comically...what..florid, maybe, and any graphic artist worth their salt can do better, sure. I know I would be cracking up saying "good one mates, but we can't really put that graphic in there". But I'm having trouble getting all torches and pitchforks.

Now, if they're using AI to mangle the *results* then I'll bring the lighter fluid and matches. And the reviewers are clearly worthless. But I'm clearly missing something because I get the sense we've turned "dorks didn't learn the tools and now look foolish" into "everything that AI has touched must be cleansed in the Butlerian jihad" (and that boat has sailed, kids).
posted by kjs3 at 3:07 PM on February 16 [1 favorite]


so THAT was the secret of N.I.M.H!
posted by Dr. Twist at 3:10 PM on February 16 [33 favorites]


Now, if they're using AI to mangle the *results* then I'll bring the lighter fluid and matches. And the reviewers are clearly worthless.

Figures in a scientific manuscript are part of the argument. They’re not just window dressing.

The problem here is not the size of the genitalia, it is that at best, the labels are badly wrong. I don’t know enough about the biological pathways in question to comment on the rest of Figure 1, but looking at the obviously fraudulent Figure 2 in which nearly every entity is one of two entities, sometimes misspelled, I see no reason to believe that the scientific argument being made by Figure 1 is accurate either. This story is not primarily a dick story.
posted by eirias at 3:15 PM on February 16 [18 favorites]


This story is not primarily a dick story.

This was the explanation I was looking for. Not just silly, silly and wrong. Thx.
posted by kjs3 at 3:18 PM on February 16 [3 favorites]


Never forget!
posted by jeffburdges at 3:29 PM on February 16 [3 favorites]


p >> 0.05
posted by dephlogisticated at 3:31 PM on February 16 [6 favorites]


MetaFilter: this whole thing is such an insult to people who actually work hard and try their best.
posted by ZaphodB at 3:35 PM on February 16 [11 favorites]


Hell is empty and the devils are all here.
posted by Ryvar at 3:41 PM on February 16 [2 favorites]


Demanding that a paper about rat genitalia have figures made by someone who knows a little about rats, genitalia or how three-dimensional space works is exclusionary gate keeping.
posted by thatwhichfalls at 3:51 PM on February 16 [14 favorites]


Come for the dicks, stay for the systemic fraud!

wait: isn't this the campaign slogan of the republican party?
posted by lalochezia at 4:09 PM on February 16 [12 favorites]


The Dissilced is so enormouse it leaves the page!
posted by amanda at 4:21 PM on February 16 [2 favorites]


I got yer retraction right here....
posted by amanda at 4:21 PM on February 16


It’s very annoying to have to think of a scientific paper as NSFW.

I have literally published a paper with a photograph of bats performing homosexual fellatio on it, and I found that rat alarmingly off-putting. It's like the image generator thought rats were just full of testicles from the rib cage on down.
posted by sciatrix at 4:39 PM on February 16 [32 favorites]


This is genuinely horrifying, actually. This is a journal with an impact factor of 6.3, which puts it in the top 5% of most-cited journals.

Figures in articles are not there for aesthetic purposes. They are not jazzy banners dressing up your drab webpage. They are there to communicate information. The information in these figures is complete gibberish. Nothing is being communicated. Or rather, completely incorrect things are being communicated. I don’t have to know anything about rat penises to know that—beyond the obvious nonsense of the text, Figures 2 and 3 are completely implausible as scientific diagrams. Figure 2 is full of extraneous details that are both *labeled* and *undescribed*—JK, STAT, SA, etc. all show up in the diagram and nowhere in the paper. Figure 3 repeatedly labels the same part of the image as different things—the green bit in D is simultaneously three different bits of gibberish I won’t bother to type up. Even if it were perfect English, it would be obviously incorrect as a diagram.

These figures are referred to throughout the paper as illustrates of concepts such as “interleukin 4 activates STAT6 molecules in T cells, which then bind to enhancer regions in T cell nuclei, prompting phenotypic changes and differentiation into helper T 2 cells.” If the diagram is nonsense, and the researchers could not tell at a glance that it was before including it, how high is the likelihood that the sentence itself is also nonsense (or at the very least not understood or supported by what the researchers actually know)?

This is a review article. There are no “results.” It is supposed to synthesize information about what we already know about a topic. This, in my eyes, makes it even more egregious. If the diagrams are gibberish, then the conclusions almost certainly are too. Because otherwise you would be able to tell, upon image generation, how very incorrect it is.
posted by brook horse at 4:54 PM on February 16 [35 favorites]




I have literally published a paper with a photograph of bats performing homosexual fellatio

and yet i get in trouble when i do this in my illustration of thermodynamic principles in my lab manual!

what better picture than for "a dire bat dick expansion"
posted by lalochezia at 4:56 PM on February 16 [11 favorites]


Christ, what a asspeehole!
posted by GenjiandProust at 5:02 PM on February 16 [1 favorite]


Setting aside the jokes, IMO this is bad enough that the entire publication should be burned to the ground. Nobody reputable should ever publish there again, and I swear to god if I were a science-type person who ever HAD published there, I'd put a little asterisk next to the citation with a footnote saying "Sorry, this was before I knew about the rat thing"

...meanwhile, any OTHER sane publication should more or less blacklist those authors. The way you enforce compliance is by destroying careers.

(then again, I do always err on the side of swift brutality)
posted by aramaic at 5:06 PM on February 16 [6 favorites]


It's like the image generator thought rats were just full of testicles from the rib cage on down.

Have I got a funny story for you about domesticated runner ducks!
posted by stet at 5:47 PM on February 16 [2 favorites]


My favorite part is the label that just reads "rat."
posted by Well I never at 5:52 PM on February 16 [6 favorites]


░D░I░ƨ░L░O░C░T░T░A░L░ ░I░N░ ░B░I░O░
posted by meehawl at 6:36 PM on February 16 [4 favorites]


looking at the illustration I can't help but think "seems like a not insane attempt to represent something small relative to the attached animal when you want focus on that thing and show a lot of detail

Not trying to be mean here, but the labels are largely gibberish. Clearly gibberish. They don't even conform well to the formation of technical jargon (i.e., they're not even a "good" AI hallucination)--there are many words they could've used in many ways and I, an expert in neither biology nor rat dicks, wouldn't have necessarily known, but when I see stand-alone phrases like "dissilced" and "testtomcels," I mean, I'm not a total illiterate.

There are ways to signal that you're depicting blown-up cross-sections, of which this is not one.
posted by praemunire at 6:49 PM on February 16 [7 favorites]


They should've run the article through a turbo encabulator first - or maybe even an experimental HyperEncabulator.
posted by adamg at 7:18 PM on February 16 [2 favorites]


I can't tell from the article... What was the purpose of this paper? Clearly not any research. Is it so the people who wrote it can add it to the list of things they've published, or were they hoping to expose the publishers, like that "Get Me Off Your Fucking Mailing List" paper?
posted by Zumbador at 7:33 PM on February 16 [2 favorites]


The rat’s expression of mild wonder tho
posted by chesty_a_arthur at 7:59 PM on February 16 [10 favorites]


I do art, including scientific diagrams. I have 2 degrees in biology and worked hard to draw well enough to make it part of my income. I really hate that some portion of my stuff might be going into this--a thing that flies in the face of any scientific integrity, especially since the authors I've worked with go over my images with a fine tooth comb to ensure accuracy. They could have paid someone like me, but they opted for a cheap route to the detriment of their reputation.

There are some VERY wonky diagrams out there. I've even made some at an author's request. But usually the wonkiness or exaggeration has a purpose, NOT blatantly wrong labels or anatomy. Look up "Cortical Homunculus" for some examples of that.

What's going on in the mouse dick is just plain wrong. What's all the blue shit? There's no legible labels. Why does it have so many testes(?) floating around in there? Why is the cross section of the dick merging with one of the testes? WTF am I supposed to learn from this image? WHY IS A SPOON SCOOPING UP GINORMOUS VEINY GLOBULES IN A PETRI DISH? STEM (Sternm) CELLS DON'T LOOK LIKE THAT. not even an abstract, exaggerated representation of stem cells ever looks like that.

The signalling pathways are illegible. Scientists don't just like to decorate their papers with little circles and squares for the a e s t h e t i c. It's supposed to be a map. Do you drive with Google Maps having a random assembly of lines and gibberish and think it's just "travel decoration" on your phone?

I really cannot stress how much I hate everything about "Image Remixers" and the marketing schemes they come up with to make people think it's "thinking" or "inspired" or whatever. But it certainly does bring out the stupidity of everyone getting in on that bubble.
posted by picklenickle at 8:33 PM on February 16 [29 favorites]


Maybe it is a little too senctolic, even if they dissilced the hell out of the testtomcels on that retat dck. I never could recetein any Sinkeclerisms, myself.
posted by snuffleupagus at 9:02 PM on February 16 [14 favorites]


Part of me actually finds it comforting that this paper will surely be scraped up and used to train countless other "AI" models, so that these future models, when asked to describe a rat, will confidently assert something like "rats have huge dissilceds and are mostly testicle from the rib cage down."
posted by dsword at 9:58 PM on February 16 [8 favorites]


That is the potential future of AI, though, right? Cheap AI starves out most actual human production and just feeds on itself in an endless nonsense replication loop.
posted by praemunire at 12:16 AM on February 17 [4 favorites]


"I've found the perfect search term to exclude AI art from image search results! just put before:2022"

i used to have a recurring fantasy a couple years ago that was like "imagine if it became impossible to record new data and 200 years later all of human culture was still just based off a snapshot of just before that event"… well.. @cirkelnio

i feel like the reason it's not Obviously Unsettling to many is because the guaranteed-to-be-real data is just a few years old, but this is obviously going to slow down the evolution of culture when most future commercial art will just be an average of whatever came before it @cirkelnio
posted by jeffburdges at 12:41 AM on February 17 [4 favorites]


It's a paper about rat genitalia. Why does the diagram include a dickcissel?!
posted by polytope subirb enby-of-piano-dice at 2:11 AM on February 17 [1 favorite]


If you are going to read about a created image of animal genitals this week on the internet make it this one as there are far worse ones going on.

At least this one will fade away next week.
posted by rough ashlar at 6:13 AM on February 17 [1 favorite]


Fraud spreads in Science by Sabine Hossenfelder
posted by jeffburdges at 1:36 PM on February 18 [1 favorite]


Frontiers statement concerning the article
Thanks to the crowdsourcing dynamic of open science, we promptly acted upon the community feedback on the AI-generated figures in the article "Cellular functions of spermatogonial stem cells in relation to JAK/STAT signaling pathway", published on 13 February 2024. Frontiers has now retracted and removed the article from the databases to protect the integrity of the scientific record.

Our investigation revealed that one of the reviewers raised valid concerns about the figures and requested author revisions. The authors failed to respond to these requests. We are investigating how our processes failed to act on the lack of author compliance with the reviewers' requirements. We sincerely apologize to the scientific community for this mistake and thank our readers who quickly brought this to our attention.
Hilarious they lead with a Statement of Great Success about crowdsourcing. Before getting around to admitting their peer review process completely broke.

Does this journal charge a subscription fee to libraries? I think not, but they do charge $3300 to get an article like this published.
posted by Nelson at 2:52 PM on February 18 [1 favorite]


I think not, but they do charge $3300 to get an article like this published.

they charge by the inch, don't they?
posted by pyramid termite at 4:40 PM on February 18 [1 favorite]




« Older Random Ex-President Hit With $364 million dollar...   |   Stikkan Newer »


This thread has been archived and is closed to new comments