A weird phrase is plaguing scientific papers | Hacker News Reader

A weird phrase is plaguing scientific papers

theconversation.com

81 pointsgreyadept37 comments

> For instance, observers have noted the rise of “tortured phrases” used to evade automated integrity software, such as “counterfeit consciousness” instead of “artificial intelligence”.

This has been a thing for years in the social media / podcast space. Unalive, bubbling yourself, vitamins, pdf files etc. in lieu of forbidden words because of the holy advertiser money.

what does "bubbling yourself" mean and cover for? I've not heard of it before. Sounds kind of fun and harmless. I did a search and it seems to refer to the act of putting oneself in a bubble, that is surrounding oneself intentionally in a social network group identity bubble. (A kind of psychological defence method like repression which in the long term isn't good). What forbidden words does it replace?

It's used by some for "killing yourself". Not as fun and harmless.

Thanks and that search engines are advertising vehicles reinforce your comment! They don't want people to find what people are using that term for.

Edits: Oh and the harmless and fun is the disguise!

Google's AI told me it's "looking for or being interested in Pastorelli Yourself Ys Pure Bubble floor tiles".

All in all I shouldn't be surprised, considering advertising is Google's business.

It comes from that Wii mario game where in multiplayer you’d jump to death and bubble up and it would auto follow the still living (best) player to finish the level.

Wait, hold on.

In that Wii Mario game players could “bubble themselves” to avoid death. And then shake their controller to bring their bubble closer to an active player, who could pop the bubble and turn the “bubbled” player back into an active player. And if all players were “bubbled” then you’d go back to the world selection screen without any player losing a life - you’d only lose any powerups you had.

So is this the correct source for the terminology, or did nobody using it actually play the game, or???

This is why you hear political podcasts talk about “the Austrian painter” or “midcentury Germans” and not the common names by which those things are known. Gotta keep that sweet YouTube monetisation.

It's strange that societies discover every 100 years or so how profanities come to existence.

This is an interesting emergent trend though because it's designed to evade automated content moderation. The whole point of "PDF file" is that it's perfectly innocuous without an understanding of the surrounding context. With real-time audio transcription increasingly common, even livestreamers have to do this to avoid getting hit with automated bans or deprioritization.

What is PDF file a euphemism for? That was the only one in the original post I couldn’t figure out.

Porn?

a nonce

Ohhhhhh.

There is a long history of euphemisms taking over - one day we may say "unalived" as a perfectly normal replacement for "killed", just as we don't call bears "arkto" or "ursa" any more, we just euphemistically call them "brown things", i.e. "bero" in protogermanic hence "bear" in English.

> This is why you hear political podcasts talk about “the Austrian painter” or “midcentury Germans” and not the common names by which those things are known. Gotta keep that sweet YouTube monetisation.

I've never heard anyone on youtube or on a podcast use terms that referred to the Nazis without directly calling them Nazis, and it hasn't had an effect on their monetization. I wonder what kind of media you're consuming that put you in contact with people who would use euphemisms for this sort of thing?

Exactly which platforms will penalise namedropping Hitler varies and it varies by circumstances, but this is a real thing as far as I can tell, on tiktok and also on YouTube videos algorithmically flagged as "intended for children" (regardless of whether they are or not).

> on YouTube videos algorithmically flagged as "intended for children" (regardless of whether they are or not).

There's a huge button where you can set a video as being child-friendly or not. I don't understand :P

I've seen youtube videos with the nazi swastika humorously censored with a $ monetisation sign. Are all those people doing this unnecessarily?

"seggs", "unalived". ugh.

It just goes to show that you can't easily change human behavior. Humans, and reality in general, will route around whatever petty schemes the ruler of the day dictate.

There is an epidemic of scientific fraud in Iranian academic circles. Not that academic fraud is isolated to Iran but it’s got a higher prevalence there

https://www.iranintl.com/en/202306138216

Similar to how cartographers embed false information into maps to catch plagiarists, I expect a deliberate practice such as this to proliferate to catch specific sources used by AI.

https://en.wikipedia.org/wiki/Fictitious_entry

The model makers can also embed such things to prove if someone copied the model or perhaps that they trained on their model's outputs.

Would also be fun to make a model that has no idea that dropbears aren't real, but thinks that Bigfoot is real, but is taboo to talk about.

I understand why AI would find and use the weird phrase (vegetative electron microscopy) if it's used on the Internet. But I'm confused about the term being used in scientific papers? Does that mean the scientific papers were written by (or with the help of) AI systems? And the folks writing the scientific papers didn't proof read their papers?

> Does that mean the scientific papers were written by (or with the help of) AI systems? And the folks writing the scientific papers didn't proof read their papers?

That seems to be exactly the implication, which is extremely disappointing.

Not surprising at all though.

It was digitized and translated wrong by computer systems pre-Generative AI then GenAI comes along and uses it in reference material. People either reading the original text or brainstorming via GenAI comes along across this term and judge it as authoritative. So it persists

Not only that, AI is used for translation a lot nowadays, and the trust in its output seems to be quite high. If you're not a native English speaker and aren't surrounded by many, you might have a much harder time finding such a mistake while proofreading.

I've seen students hand in auto-translated garbage 10 years ago. Nowadays auto-translation works well enough that you are at least able to read it.

Small discussion (61 points, 2 months ago, 7 comments) https://news.ycombinator.com/item?id=43022628

So papers printed in double columns are scanned as one column?

I think that’s the far greater issue, and that there would be many other issues. Maybe this case is a rare exception but surely the text would be converted from a two column format to a single body before being processed.

The joys of hindsight.

It reminds me of this from 16 years ago: https://news.ycombinator.com/item?id=18374359

I wonder how long it will be before AI regurgitates that hilarity.

But many of the existing papers with that phrase also happen to be from Iran. Interesting

https://scholar.google.com/scholar?hl=en&as_sdt=0%2C5&q=%22v...

If they only differ by a dot, this could be legitimate papers being translated poorly. I don't see what the big deal is. Is the suggestion that these journal articles just AI garbage? I thought the editorial boards were supposed to be able to put a stop to that.

I'm assuming they are using LLMs for translation, which makes this mistake as it already knows about "vegetative electron microscopy".

Asked an Iranian-German electron microscopist that I know.

Scanning = robeshi

Vegetative = royeshi

Probably just a typo. Scanning electron microscopes (SEM) are very common instruments.

this is in the article

And here I thought it would refer to "unable to replicate" or "photoshopped image"

Reply on news.ycombinator.com