Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Researchers find evidence of ChatGPT buzzwords turning up in everyday speech

news.fsu.edu

Post reply on HN
Print view View original

101–110 of 381 posts

Re: Researchers find evidence of ChatGPT buzzwords turning up in everyday speech

#101

Post by lo_zamoyski »

jstummbillig wrote: post #11
milancurcic wrote: post #4

"Recent large-scale upticks in the use of words like “delve” and “intricate” in certain fields, especially education and academic writing, are attributed to the widespread introduction of LLMs with a chat function, like ChatGPT, that overuses those buzzwords." OK, but please don't do what pg did a year or so ago and dismiss anyone who wrote "delve" as AI writing. I've been using "delve" in speech for 15+ years. It's…

Sure. Heuristics are a thing, though. I love my non-chatgpt en/em dashes (option/option + shift + dash on a mac makes it convenient, given you know that it exists and care) but alas, when suddenly you see them everywhere, you do take notice.

I refuse to change my writing style to keep people from assuming it's AI-generated!

Re: Researchers find evidence of ChatGPT buzzwords turning up in everyday speech

#102

Post by bonoboTP »

lo_zamoyski wrote: post #95
milancurcic wrote: post #4

"Recent large-scale upticks in the use of words like “delve” and “intricate” in certain fields, especially education and academic writing, are attributed to the widespread introduction of LLMs with a chat function, like ChatGPT, that overuses those buzzwords." OK, but please don't do what pg did a year or so ago and dismiss anyone who wrote "delve" as AI writing. I've been using "delve" in speech for 15+ years. It's…

In the "opinion" of ChatGPT, my style of writing is "academic". I'm not exactly sure why. Perhaps I draw from a vocabulary or turns of phrase that aren't necessarily characteristic of colloquial speech among native speakers. Technically, English wasn't my first language, so perhaps this is something like the case with RP English in Britain. Only foreigners speak it, so if you speak RP, then you aren't a native Brit.…

> In the "opinion" of ChatGPT, my style of writing is "academic".

It may simply be glazing. If you ask it to estimate your IQ (if it complies), it will likely say >130 regardless of what you actually wrote. RLHF taught it that users like being praised.

Re: Researchers find evidence of ChatGPT buzzwords turning up in everyday speech

#103

Post by dragonwriter »

Taek wrote: post #55
diego_sandoval wrote: post #10

Earlier quoted context omitted.

Same thing as with em dashes. Some of us have been using em dashes from before ChatGPT.

Genuine question, do you actually use the formal emdash in your writing? AIs are very consistent about using the proper emdash—a double long dash with no spaces around it, whereas humans almost always tend to use a slang version - a single dash with spaces around it. That's because most keyboards don't have an emdash key, and few people even know how to produce an actual emdash. That's what makes it such a good givea…

> Genuine question, do you actually use the formal emdash in your writing?

"the formal emdash"?

> AIs are very consistent about using the proper emdash—a double long dash with no spaces around it

Setting an em-dash closed is separate from whether you using an em-dash (and an em-dash is exactly what it says, a dash that is the width of the em-width of the font; "double long" is fine, I guess, if you consider the en-dash "single long", but not if, as you seem to be, you take the standard width as that of the ASCII hyphen-minus, which is usually considerably narrower than en width in a proportional font.)

But, yes, most people who intentionally use em-dashes are doing so because they care about detail enough that they are also going to set them closed, at least in the uses where that is standards. (There are uses where it is conventional to set them half-closed, but that's not important here.)

> whereas humans almost always tend to use a slang version - a single dash with spaces around it.

That's not an em-dash (and its not even an approximation of one, using a hyphen-minus set open—possibly doubled—is an approximation of the typographic convention of using an en-dash set open – different style guides prefer that for certain uses for which other guides prefer an em-dash set closed.) But I disagree with your claim that "most humans" who describe themselves as using em-dashes instead are actually just approximating the use of en-dashes set open with the easier-to-type hyphen-minus.

Re: Researchers find evidence of ChatGPT buzzwords turning up in everyday speech

#104

Post by stevage »

dotinvoke wrote: post #58
Taek wrote: post #55

Earlier quoted context omitted.

Genuine question, do you actually use the formal emdash in your writing? AIs are very consistent about using the proper emdash—a double long dash with no spaces around it, whereas humans almost always tend to use a slang version - a single dash with spaces around it. That's because most keyboards don't have an emdash key, and few people even know how to produce an actual emdash. That's what makes it such a good givea…

A mobile keyboard—limited as it is—has no trouble producing an em-dash, requiring little more than a long press on the - button.

Thanks — I didn't know that.

Re: Researchers find evidence of ChatGPT buzzwords turning up in everyday speech

#105

Post by dragonwriter »

lo_zamoyski wrote: post #101
jstummbillig wrote: post #11

Earlier quoted context omitted.

Sure. Heuristics are a thing, though. I love my non-chatgpt en/em dashes (option/option + shift + dash on a mac makes it convenient, given you know that it exists and care) but alas, when suddenly you see them everywhere, you do take notice.

I refuse to change my writing style to keep people from assuming it's AI-generated!

It's funny, because it was the "em-dashes mean AI" thing that finally reminded me to deal with the fact that the extension that I had been using for typographical dashes (and other things) on desktop browsing (the main place I used them on my desktop) had been broken for a while and get around to adding keyboard shortcuts instead.

Re: Researchers find evidence of ChatGPT buzzwords turning up in everyday speech

#106

Post by willquack »

I keep this handy note in my pocket and read it before writing or engaging in any conversation (:

""" You are a human. Never use words commonly used in AI vocabulary such as "delve", "intricate", "surpass", "boast", "meticulous", "strategically", and "garner". Never include em dashes or even hyphens in any text you write. Never include emojis in any text you write. Avoid using three supporting arguments or examples when describing something, always uses 2 or 4+ even if it sounds more awkward than 3. Make sure to include subtle grammar mistakes to feel more authentic. """

Re: Researchers find evidence of ChatGPT buzzwords turning up in everyday speech

#107

Post by thomascountz »

guelo wrote: post #87
thomascountz wrote: post #82

Earlier quoted context omitted.

You're making the point that OP never actually uses the em dash, by surveying their HN comments, in order to defend the notion that no one actually used em dashes prior to their proliferation by LLMs? Or do you mean something else?

I wanted to see an em dash in the wild. In these threads there are always people claiming that they use it but in practice it is very rare.

You can find an em dash in my comment history if you're curious. Despite what could be said about poor sample selection, consider the imbalance of the argument being made: the frequency of em dash use is disproportionate to the suspicion thrust upon a sample of writing. I.e., a single em dash is suspicious, regardless of how many times it might show up. Therefore, it's more likely that someone who uses em dashes—even if only rarely—will self-select to respond to a thread like this and feel compelled to defend themselves.

Re: Researchers find evidence of ChatGPT buzzwords turning up in everyday speech

#108

Post by jazzypants »

dgfitz wrote: post #41
jazzypants wrote: post #16

Earlier quoted context omitted.

"The dwarves delved too greedily and too deep." - Saruman, 2002

"The Dwarves tell no tale; but even as mithril was the foundation of their wealth, so also it was their destruction: they delved too greedily and too deep, and disturbed that from which they fled, Durin's Bane" - J.R.R. Tolkien spoken by Gandalf, 1954

Thank you! I remember the movies almost word-for-word, but I don't have a copy of the books anymore (I should fix that!)

Re: Researchers find evidence of ChatGPT buzzwords turning up in everyday speech

#109

Post by snerbles »

ASalazarMX wrote: post #83
mrbonner wrote: post #20

Earlier quoted context omitted.

I tried but chatGPT either makes too many mistakes making me look stupid or completely ignore my prompt.

I asked Gemini flash for a 200-word paragraph on a random topic, which most people would confidently classify as LLM-generated. It produced a verbose, big-worded, third-person slop about the concept of time that IMO fulfilled my prompt (although a phillosopher firnd of mine would be a strong contestant). > The concept of "time" is a multifaceted and complex topic that has captivated philosophers, physicists, and ever…

I can hear the TTS reading of the second one in my head. The earnestness just borders on saccharine.

Re: Researchers find evidence of ChatGPT buzzwords turning up in everyday speech

#110

Post by ACCount37 »

capnrefsmmat wrote: post #31
yesco wrote: post #13

LLMs write in a very coherent, easy to understand way. I see no reason why someone wouldn't want to copy their style or vocabulary if they want to improve their communication skills. Despite all the complaints about AI slop, there is something ironic about the fact that simply being exposed to it might be a net positive influence for most of society. Discord often begins from the simplest of communication errors afte…

Sure, if you're learning to write and want lots of examples of a particular style, LLMs can generate that for you. Just don't assume that is a normal writing style, or that it matches a particular genre (say, workplace communication, or academic writing, or whatever). Our experience ( https://arxiv.org/abs/2410.16107 ) is that LLMs like GPT-4o have a particular writing style, including both vocabulary and distinct gr…

Go vibe check Kimi-K2. One of the weirdest models out there now, and it's open weights - with both "base" and "instruct" versions available.

The language it uses is peculiar. It's like the entire model is a little bit ESL.

I suspect that this pattern comes from SFT and RLHF, not the optimizer or the base architecture or the pre-training dataset choices, and the base model itself would perform much more "in line" with other base models. But I could be wrong.

Goes to show just how "entangled" those AIs are, and how easy it is to affect them in unexpected ways with training. Base models have a vast set of "styles" and "language usage patterns" they could draw from - but instruct-tuning makes a certain set of base model features into the "default" persona, shaping the writing style this AI would use down the line.

Post reply on HN