Scientific reports in the age of AI - This actually got published at Nature.com

[highplainsdem]

"Runctitiononal features"? "Medical fymblal"? "1 Tol Line storee"? This gets worse the longer you look at it. But it's got to be good, because it was published in Nature Scientific Reports last week: www.nature.com/articles/s41... h/t @asa.tsbalans.se

— Erik Angner (@erikangner.com) 2025-11-27T09:30:27.070Z

UPDATE: The publisher intends to retract the paper, but insists it went through two rounds of review from two independent peer reviewers: nobreakthroughs.substack.com/p/riding-the... by @jacksonwryan.com. No word on how they will make sure this sort of thing never happens again.

— Erik Angner (@erikangner.com) 2025-11-28T08:17:23.081Z

The "scientific report"

https://www.nature.com/articles/s41598-025-24662-9

haa this added to it as of today:

28 November 2025 Editor’s Note: Readers are alerted that the contents of this paper are subject to criticisms that are being considered by editors. A further editorial response will follow the resolution of these issues.

[Wounded Bear]

1. The new paradigm is minimal editorial oversight...

Articles go almost from the writer to the printer/publisher without much editing.

Weird shit gets through more often.

[highplainsdem]

5. That figure shouldn't have gotten past anyone who wasn't blind. Including the author.

[Wounded Bear]

6. No argument here...nt

[SheltieLover]

2. I'd be looking hard at the reviewers

[KT2000]

3. AI reviewers?

This is going to be fun!

[nilram]

13. "Peers" of the author, apparently

[dalton99a]

4. Famous author:

[eppur_se_muova]

12. Vocational ! College of Press ! and Publishing ! ... just the sort of place you'd expect to be doing autism research.

You might have added the emoji to your post for clarity, but it was pretty obvious.

[LudwigPastorius]

7. "...it went through two rounds of review from two independent peer reviewers."

*
Spoiler Alert*

Those peers were, themselves, AIs.

[tanyev]

8. Just the thing Bobby Brainworm can wave around as proof he was right about everything.

🤨

[Maru Kitteh]

9. Anyone looking to ANY dot-com for reliable information needs education

I mean, a website that exists to sell you stuff might have interests at hand that don’t include accurate or truthful information, right? Kinda 101.

Not shocked.

[eppur_se_muova]

10. Holy crap -- look at the "m"(?) in "frymblal" -- it's the weirdest typo ever.

"Environental" seems pretty tame beside that.

Something about this is just not cromulent.

ETA: I can't believe I didn't notice the woman's legs first thing ....

[yonder]

15. It is not cromulent because of inadequately embiggened research and review.

[tanyev]

19. 🤣

[yonder]

26. Now we're talking!

[lapfog_1]

11. from Gemini ( Google AI )

Based on the context from the web page, the phrase "Runctitiononal features" appears to be a nonsensical term that was incorrectly published in a figure within a scientific paper in Nature Scientific Reports.

The web page mentions a post by Erik Angner highlighting the phrase and questioning its validity, suggesting it is part of what he calls "AI slop."

The term is likely a typographical error or an AI-generated artifact (a "hallucination&quot that slipped past the authors, peer reviewers, and editors, as it is not a recognized or standard scientific term.

The article that contained this error is noted to be under scrutiny and the publisher intends to retract the paper, which further indicates the presence of severe flaws like this meaningless phrase.

It seems to be an example of the problems arising from "Scientific reports in the age of AI" where minimal editorial oversight allowed a poorly-generated figure with gibberish terminology to be published.

[highplainsdem]

16. Why post AI results? Gemini can get things wrong as well. Please don't post AI slop here.

[lapfog_1]

18. AI slop?

The entire internet is slop. For that matter, most of published scientific literature is slop ( much of it plagiarized ). Quite a bit of entertainment is as well ( especially music ).

AI is just another tool. Use it, don't use it. That is up to you.

So... no... I will use it when I want ( as long as I attribute the source, which I did ).

I thought it was appropriate to have one AI fact check another ( unattributed ) "AI" ( which might have been fake to make all of AI look like it can't tell bullshit from non-bullshit ).

Every time someone has posted "how stupid is AI" example, I check the query by asking another AI to solve for the same prompt. So far, the AIs I use have checked out just fine. Of course it is good to check the results yourself.

And you have to ask yourself... if AI was so bad at everything ( slop as you call it ) why have so many companies embraced using it?.. I would think they would all start losing customers and go out of business.

[highplainsdem]

21. If you use generative AI even though it's trained on stolen intellectual property, you're acting unethically.

It is not "just another tool."

Those responsible for the IP theft should receive prison sentences.

And those who are just fine with the IP theft - who'll use genAI without being forced to by their job or school - are showing a serious lack of liberal values.

[lapfog_1]

27. so the problem is not that it is "slop" but that it works really well

and some people that create the LLMs are unethical.

Much of the copyright material used to train an AI is not copied verbatim. it is more like a "book report" and not the book. Every middle school, high school, and university student that reads a book and uses it later is guilty of what goes on in AI. Only they don't look at 1200 papers or books on a subject, they look at 1 or 2 or 5 sources ( I was taught to use at least 3 but less than 10 ). Attribution is lacking... but I didn't pay for the access to the books and papers... I read them, summarized thm,, wrote my own paper on the subject, often lifting entire passages... and included footnotes with attribution.

Like I said, AI is a tool, the internet is a tool, social media is a tool... use it, don't use it, if you are an injured party, sue the bastards. But don't blame the new tool, it is not responsible for what source material was used.

[highplainsdem]

28. GenAI is always hallucinating, not reasoning, not aware of what it regurgitated. But some of its

hallucinations look more rational than others.

It's never reliable, which is why the TOS advise users of that and try to shift legal liability for plagiarism, etc., to users. That unreliability is why genAI tools have not been as widely adopted by paying customers as the AI companies had hoped, so theu're losing incredible amounts of money and we have a bubble with companies kept afloat by venture capitalists and circular financing...and more and more AI companies are saying they want the federal government to assume the financial risks, because they know it's a bubble.

If you've been following much of the news about genAI, you should be well aware that the AI companies expected to be sued, because they knew they were breaking IP laws all along. They still hoped to get away with the theft.

It isn't and never will be "just a tool." GenAI is an illegally trained and fundamentally unethical tool, and all the rationalizations in the world won't make your decision to use it, if you aren't being forced to do so, any more ethical. Karma would have you lose most of what you own - and with the thieves, or someone who benefited from the theft, trying to justify your loss and their gain.

But I wouldn't wish that kind of loss on you or anyone else, even though you're showing zero concern for all the people whose IP was stolen by AI robber barons.

[Cirsium]

37. The entire internet is slop?

What does that even mean?

[lapfog_1]

38. most of the content of the internet is not original thought

recent revelations on X show that much of the right wing magasphere leading thread contributors originated in other countries... plus a lot of "likes" and retweets from bots and click farms.

In the context of calling AI "slop"... I think the entire internet ( something that I helped invent btw ) is mostly crap.

[Cirsium]

40. Sad

I think you are right.

[UTUSN]

14. Am adding "AI sleuths" to my lifelong list of Annoying Irritants (AI) -

O.K., fine, if our internet gadflies alert us when we are sincerely taken in by a misleading AI thing. But some of us in the 78 age range post something that is INOFFENSIVE or whose content is on "our side" and some Annoying Irritant debbie-downer/website-nanny/thread-killer immediates posts "AI"! Even worse is when we post something from an entirely legitimate source (think: AP, Reuters, NPR) and a fellow Clued-out posts, "I this real? Is it AI?"

*** At my advanced age I can pinpoint milestones of great cultural changes that required a whole Paradigm-changing thing. I guess AI is now one of those. But some of these puffed-up AI Sleuths are Annoying Irritants.

[highplainsdem]

20. You've seen people asking if something from a mainstream news source is AI? Asking about it here

at DU?

As for identifying AI content - that's important. Too many people are taken in by it, or think it's harmless. It isn't.

[UTUSN]

31. Yea, at DU at least once, a month ago. The other, irritating latest one was

Last edited Fri Nov 28, 2025, 07:08 PM - Edit history (1)

a couple of hours ago.

[highplainsdem]

32. Where's the latest one, the one you find irritating?

[UTUSN]

33. The undeniable one I self deleted as soon as the sleuth found out.

My gripe is not that it is AI or that I was enticed by the verisimilitude of the content and voice. It's having to second-guess even more (proofreading used to be bad enough) and having basic effort in posting crapped on with a tone of smartypants. Here, don't want more waste on this :

https://www.democraticunderground.com/100220835923

[highplainsdem]

34. UTUSN, you just uncovered more stuff being done by the person creating fake Obama videos earlier!

Wow! The YouTube channel that person had for the fake Obama videos has been removed, apparently, but it had a disclaimer similar to the one you posted in reply 2 here - and notice that in the disclaimer you copied, for the fake Kimmel video, the disclaimer earlier today, it refers to Gavin Newsom, not Jimmy Kimmel.

Here's that disclaimer you copied to DU, the way it still read on YouTube when you posted reply 2:

https://www.democraticunderground.com/100220835923

UTUSN
2. If it is, I'll delete this, but am getting tired of the "AI" catchall thing.
12:20 PM
I'll grant that the lips don't match, but the voice and the content is very KIMMEL-like. If the content is real KIMMEL, won't delete.

Here's the description from the link:
********QUOTE*****

America Is in the Heart T1

Description
⚠️ Disclaimer: This is a fan-made channel, and its content is not affiliated with Gavin Newsom or his companies. The videos are inspired by Gavin Newsom's public statements and ideas for educational and motivational purposes only, using a synthesized voice that does not belong to Gavin Newsom. We use visual lip-syncing and dubbed narration to match the spoken words with on-screen footage, purely to enhance clarity, create a cinematic experience, and make the content more engaging for viewers. Our aim is to amplify the original message by making it easier to understand for the end consumer, helping us reach and educate more people with Gavin Newsom's valuable perspectives. We also make the messages of Gavin Newsom more accessible to people who are deaf or hard of hearing by applying professional transcription to the majority of our videos. We share his visionary ideas in a respectful and inspiring manner, without any intent to mislead.
More info
www.youtube.com/@AmericaIsintheHeartT1

*********UNQUOTE*****

This makes me wonder if whoever has that YouTube channel is also doing deepfakes of Newsom on another channel.

After you copied that disclaimer to DU, whoever has that channel changed Gavin Newsom to Jimmy Kimmel, without changing a word otherwise. This is the current disclaimer:

America Is in the Heart T1

Description
⚠️ Disclaimer: This is a fan-made channel, and its content is not affiliated with Jimmy Kimmel or his companies. The videos are inspired by Jimmy Kimmel's public statements and ideas for educational and motivational purposes only, using a synthesized voice that does not belong to Jimmy Kimmel. We use visual lip-syncing and dubbed narration to match the spoken words with on-screen footage, purely to enhance clarity, create a cinematic experience, and make the content more engaging for viewers. Our aim is to amplify the original message by making it easier to understand for the end consumer, helping us reach and educate more people with Jimmy Kimmel's valuable perspectives. We also make the messages of Jimmy Kimmel more accessible to people who are deaf or hard of hearing by applying professional transcription to the majority of our videos. We share his visionary ideas in a respectful and inspiring manner, without any intent to mislead.

This is the disclaimer that was used a couple of weeks ago for a channel with similar fakes of Obama. Note the way the disclaimer starts, and some similar though not identical wording. I'd copied the disclaimer into a reply in a thread about one of thw videos.

https://www.democraticunderground.com/100220808654

⚠️Disclaimer:
This is an independent, fan-created channel and is not affiliated with Barack Obama, his campaign, or any political organization.

Our content draws on public speeches, verified reporting, and historical analysis, presented for educational and analytical purposes only.

We use voiceover narration, storytelling, and dramatized reenactments to make complex political news more accessible, emotionally engaging, and thought-provoking. All visuals and voices are original productions created by us, and do not feature or portray Barack Obama himself.

Our mission is to promote civic understanding and thoughtful dialogue while maintaining respect for all public figures. We are committed to responsible sharing of public information and to upholding fair-use principles and journalistic integrity.

Both the Newsom/Kimmel disclaimers and the Obama disclaimer refer to making the messages "more accessible" and to doing this for "educational" purposes.

God only knows how many YouTube channels with deepfakes of famous people this YouTuber might have set up.

[UTUSN]

35. You have *way* *far* deeper grip on the whole subject than I will ever have - or *want*!

What is my superficial gripe is that this KIMMEL piece was so entirely convincing and *not* misrepresenting KIMMEL content that I've heard from him in pieces - that I *LOVE*. After I've been alerted to the AI angle of it, 1-Why doesn't or hasn't KIMMEL (and the rest) complained about their being appropriated? 2-The content seems *favorable* on "our side," so what category of (criminality?) or misrepresentation does this fit in?

I mean, the content is TRUE, what I want to hear. After the AI Sleuth tipped off that it was AI and I went back to look at the video I couls see the lips mismatching the voice, and one of the Comments in the video said that, too.

Congrats on your savvy.

[highplainsdem]

36. There's so much of this deepfake type garbage on YouTube that they'd need fulltime staff to hunt

for it and send cease-and-desist letters.

I'm glad the channel.for the Obama deepfakes is gone, but I don't know if Obama's people complained, or if YouTube got so many complaints from others that they felt they had to remove that channel.

I'm so tired of all the ways AI is being misused. Politicians and celebrities shouldn't have to worry about AI users putting words in their mouths.

[Lettuce Be]

17. Just pathetic -- peer reviewed my ass

Seriously, they can no longer be taken seriously if their response is, "Well, it was reviewed!" Nice try, but no.

[haele]

22. Went through two peer reviews?

Methinks the reviewers either didn't review the document, or used AI to just look at the prose and assumed any figures were generated with actual data, instead of through AI.
Anyway, what was/were the author(s) thinking? Did they even try to have a human being look at the paper before publishing it?
I'd at least read my own "work" before submitting it, just to ensure I got it down right and wouldn't be surprised by any questions someone might ask me about it.

[struggle4progress]

23. "The primary objectives of this research are ... To develop an explainable AI framework using TabPFNMix and SHAP for

autism diagnosis: This involves leveraging the TabPFNMix regressor, a state-of-the-art ML model designed for tabular data, to improve the accuracy of ASD diagnosis. The framework integrates SHAP to provide interpretable explanations for the model’s decisions"

This might even lead to self-driving cars that shriek "Quiet, Piggy!" as they crash through crowds of protestors!

[hlthe2b]

24. I am running out of "disgusted" synonyms to really describe my reaction to this.

Not that many had not warned against it, but it was too late, and too many slept through the "neon signs" of the coming crisis.

[LAS14]

25. Just this morning I read about people inserting typos and mixed fonts...

...into their writing in order to avoid getting labeled as AI generated. Might that be going on here?

[Initech]

29. Cloudflare got shut down for 10 hours last week because of one typo in an executable file.

[Initech]

30. It's crazy how fast we went from 0 to 60 on AI.

[jfz9580m]

39. This is tangential to this thread

But as I was infuriated by heavy traffic on the street outside, I was thinking about the use of big data and ai riding this wave, with bullshit “studies” in tow.

I needed to rant in an ai thread so forgive me highplainsdem as it is only tangentially related best case to your op.

One infuriating thing I suspect is that with the various deregulated real world experiments ongoing in Ai and big data (which clearly encroach into the real world) they try to build junk like this:
https://www.noemamag.com/are-we-accidentally-building-a-planetary-brain/

Obviously left or right they would kick out environmentalists, scientists, journalists or regulators who don’t go just along with stuff. O They also have the most bogus types of inhouse “critics” (think the FB oversight board).

It is bound to be the most corrupt, sexist and sleazy space grab imaginable. They usually do just about anything in the most anti-democratic exploitative way imaginable the world over.

It always follows the same pattern wrt any scientist/journalist or regulator who doesn’t play ball: Timnit Gebru was fired by Google; Maria Ressa was persecuted; Madhav Gadgil’s “Gadgil report” was tossed aside by the corrupt communist CM Pinarayi Vijayan of the state of Kerala. That last is an example of how ecologists or evolutionary biologists who are not bought and paid for by corrupt business interests are thwarted when environmental laws and wildlife rhythms are messed up by poorly regulated technologists..
And with the surveillance state, the implicit threat of the use of any triviality (e.g.: marijuana laws) to harass complainants is ever present. I am so glad on that head that Duterte was sent to The Hague.

They want to build these things using big data, dubious ai and deregulated emerging technologies. They won’t stand for regulation or scientific methods in any serious sense.

This guy seems like a crank (not that I should be attacking anyone else on that head ) and worse he talks to irredeemable creeps like Roger MacNamee, but even guys like this are bashing this ai push:

https://www.the-geyser.com/ai-defends-the-status-quo/

And anything “open” sold by Google or its creepy small fellow travellers would be bilge.
(Tristan Harris, MacNamee etc define astroturf unlike say the always awesome, universal hater Yasha Levine. Zitron is also trustworthy. Haters are the only people I listen to anymore.)