A paper in the journal Pharmaceutics, Inhaled Medicines for Targeting Non-Small Cell Lung Cancer by Arwa Omar Al Khatib, Mohamed El-Tanani, and Hisham Al-Obaidi, might be good science, but it’s so far outside my field that I’m not equipped to judge. There are little things that I would have taken a red pen to — the first sentence of the abstract starts, “Throughout the years…” which is vague and clumsy and useless, but OK, we can’t expect literary excellence in everything some nerd writes — which put me on edge, but nothing to obviously indict it as a poor bit of work. That is, until I saw the figure.
Figure 1. Examples of current drug delivery strategies for the treatment of NSCLC.
Wow. What a glorious splortch of utterly useless visual noise. Is the whole paper an equivalent pile of machine-processed nonsense?
I really am wondering what the authors were thinking. Why would you insert a vivid technicolor example of obviously meaningless garbage into the middle of your work? Because you can? It really calls into question the validity of the text, and it makes the journal look like a rinky-tink circus sideshow. Where were the reviewers? The editors? Non small cell lung cancer is a serious concern — imagine a patient researching their disease, and seeing that doctors see their organs as exploding balls of orange and blue, like the contents of a SF movie poster.
This is an AI-generated image, isn’t it? It just has that feel to it.
@ ^
Yes that’s PZs point. He’s not a fan of “AI”.
Are my lungs really that garish?
Unless Elon Musk develops that brain transplant thingand cyborgises you so you can see your own lung s befre you I’m guessing we’ll never know.. evidnece for medical cadavers and autopsies almost certaionly says no.
Although actually I guess with what they can do with tiny cameras and medical whatsmajigs these days ..maybe ?
I don’t remember seeing credible, legitimate science books showing stuff like this. This is pure AI here.
Clarity fix (memo to self – stop trying to post quickly in TV ad breaks) :
If one does have lungs that garish I imagine it wouldn’t be terribly healthy and some sign of some awful medical condition. – perhaps something involving bioluminescent fungal spores infecting and badly inflaming lung tissues?
“What a glorious splortch of utterly useless visual noise.”
The picture has to be AI generated. I have a feeling we will be seeing this more often, until maybe this is the only way they produce pictures. This picture is more visually stimulating then informative, even though the individual sections make sense. Reminds me of my college English teacher telling us we better not hand in any papers dripping in Purple Prose.
I’ve seen human lungs. The anatomy lab next door sometimes leaves them laid out on a table in the prep room (also all the other organs you can imagine.)
They aren’t particularly colorful, or spherical, or even shiny.
Individual sections make sense? I know what a tyrosine kinase is and does, and how it can be important in cancer therapy, but I have no idea what is illustrated in the lower right corner, for instance. None of it communicates anything useful.
The text is a lot more lucid than I’d expect from machine-learning generators (ML image generators usually make text that looks like this). The style looks like an ill-informed choice, and the juxtapositions are kind of wonky, but there are aspects of it that could be the result of misguided human design (disclaimer: I know nothing about drug delivery and to what extent these pictures and labels are total nonsense. I verified that tyrosine kinase inhibitors have a structure which vaguely resembles that in the picture). The one part which absolutely looks like a ML artifact to me, though, is that pill dissolving into a puddle of goo in the left-middle top.
That’s where the danger in these illustrations lies. People try to interpret the image in a way that makes sense to them, and force fit an interpretation that doesn’t work. Tyrosine kinase inhibitors are generally small molecules with a couple of cyclic structures (examples), and they look absolutely nothing like that orange and blue exploding blob in the figure. I struggle to relate anything in the image to anything in molecular biology or cancer treatment — nothing is relevant or of any utility.
That diagram destroys any respect for the authors or the editors. The text is also AI generated bullshit. It’s clear very early on.
Nope. Ventilation does not move air ‘from the blood’. The AI here does not understand the difference between ventilation, diffusion, and gas exchange. It just cobbles together sentences from words that commonly cluster around those terms.
Conceptually muddled. Elasticity does indeed promote rebound to resting volume and reduces the work of breathing, but if lung elasticity keeps increasing, it becomes fibrotic disease rather than an enhancement because the lung rebounds to rest state so powerfully that it’s hard to breathe in.
It has all the hallmarks of AI: lack of conceptual understanding, hodgepodges of facts pulled from the literature that are mostly correct but have no relevance to the topic at hand, and unstructured paragraphs that leap from factoid to factoid with tenuous logical connection.
This is just a slurry of observations from other published papers. It has no internal coherence, has scattered the observations on smoking into unlinked sentences and clauses, uses wildly inappropriate qualifiers that sound vaguely academic but are meaningless in context (what makes the link between smoking and cancer ‘intrinsic’?), describes ‘exposure to harmful substances’ as a separate entity to ‘environmental pollution’ because it can’t do simple categorical reasoning, and is unaware that that HIV increasing cancer risk is in fact unsurprising (HIV reduces immune function and increases the risk of many cancers). The phrase ‘certain genetic markers like the CYP1A1 gene variant’ demonstrates the lack of conceptual processing going on here. CYP1A1 is a gene, with many variants, several of which are associated with increased lung cancer risk. The AI doesn’t know the difference between a gene, a gene variant, and a marker, it just found these words in statistical clusters and threw them together.
The paper gets confused later by the difference between natural and synthetic surfactants. And for a paper on inhaled lung medications, it seems an extraordinary oversight not to even mention therapeutic surfactant for premature neonates, a treatment that has been an important medical intervention in common use since the 1980s.
I didn’t read all of it. This was more than enough.
@9– that diagram doesn’t look like a tyrosine kinase to me.
molecular model
diagrams…from more reliable…sources. Note the conspicuous absence of any trachea.
I’ve seen a pair of human lungs at a food market no less (historic Lexington Market in Baltimore) as part of an anti-smoking demonstration. Even got to touch them with gloves on. Of course they looked nothing like what’s pictured here.
I recall a Ray Bradbury story where a kid realises the neighbor is not human because his insides are NOT like what you expect (most of you may have read it). This is an illustration of those organs.
🙂
Can you please add an amendment to your constitution about ‘no AI in textbook creation” except maybe for designing diagrams.
“The Man Upstairs.” The man is a vampire, so the kid was able to cut him up during the day, and bring down all the weird body parts without any response.
He dumped his piggy bank full of silver dimes into the still squirming hollowed out corpse.
I remember it well because my grandparents had an upstairs room (they didn’t have any boarders, though) with a stained glass window like the one in the story. I knew what I was going to do if we were invaded by vampires!
This is how anti science gets fostered.
Peer review in journals needs to get stronger. Junk journals need to be called out.