Submitted 2 days ago by xkcdbot@lemmy.world [bot] to xkcd@lemmy.world
https://imgs.xkcd.com/comics/disclaimer.png
Title text:
You say no human would reply to a forum thread about Tom Bombadil by writing and editing hundreds of words of text, complete with formatting, fancy punctuation, and two separate uses of the word ‘delve’. Unfortunately for both of us, you are wrong.
Transcript:
Transcript will show once it’s been added to explainxkcd.com
Source: xkcd.com/3126/
Yep ChatGPT must have learned from people like me, because:
Claim:
- I use bullet points in every post
Fact Check:
Out of your 36 comments, this is the only one with bullet points. That’s only 2.7% of your comments. One other has an enumeration, but an enumeration is not bullet points.
Additionally, you have one post, but that also doesn’t use bullet points. 0% of your actual posts use bullet points.
Conclusion: Claim is FALSE. Ziffy-fa-Jazz-KZone-Sweek’em does not use bullet points in every post.
Nice :) I like that you did that! But to clarify some things:
Chat gpburn
Fact: we don’t know and can’t come to a definite conclusion
Hey, that’s a really well-formatted commemt! If it weren’t for all my typos, people would probably suspect me too – but hey at least now I have an excuse not to edit anymore.
(Edited: 2 minutes ago)
Emoji, title, bullet points, repeat
Can you elaborate on the second point?
Well now they literally cannot, since you asked!
I’ve had a handful of people think I was AI too because of my fancy word choice and mediocre knowledge of punctuation. My writing voice is ubfortunately devoid of emotion most of the time, and when it isn’t, it tends to fall into the matter-of-factly category. I try to include punctuation or breaks to be a little more similar to my speaking such as “well, uh, you wouldn’t…” or “I- Yeah, I don’t know.” I think my personal favourite I do do is “I think it was… two…? weeks ago?”
And yet I still get dubbed AI because I periodically use the word plethora…
i found splitting up paragraphs into many, small, newlined sentences help a lot. it seems to be mainly big paragraphs that give people uncanny feeling.
or alternatively join me. in not capitalizing letters anymore.
it damn well bloody helps to include a fucking shitload of swearing, e-fucking-specially if it’s a mite creative too.
Okay. Here’s a detailed response to BombaFan486’s post, including three separate (but natural) uses of the word ‘revelatory’, ending with the disclaimer, “Not ChatGPT. I’m just like this.”
Is there anything else I can do for you?
There really is an xkcd for everything.
Me. I like formatting and using em/en dashes and semicolons.
Imitating LLM to piss off AI haters is just next level trolling. As everyone knows, trolling is a art.
Seeing any em dash in the wild makes me immediately suspicious
I loooove using m dashes. They’re just so—expressive.
Yeah, I’m 42 yrs old. A long time ago I sacrificed what lvl of English I did know in order to use “lol” - I’m not giving up em dashes.
There’s always one good litmus test, if you can ask someone live before they have a chance to do a Google search: ask what the alt code is for an em dash.
If they don’t immediately answer 0151, they’re full of shit.
I feel like this is more likely to lead you astray than anything. An LLM bot will immediately know the alt code, while a real person will only know it if they use Windows. Lots of people use Linux, or mobile keyboards.
I always just google ‘em dash’ then ctrl+c it.
meh - I just use a dash and move on.
for me it’s just altgr+caps followed by tapping - thrice, way better use of the caps lock key. thank u xmodmap
Opt shift hyphen. What’s an alt code? 🍏
Or if they start explaining what a compose key is; also human
What if they don’t know the code and just use a keyshift instead?
You just have say fuck a lot…
But I’m pretty sure any explanation of Bombadil less than 300 words would fail the Turing test
That is an excellent point! Use of the word “fuck” in online conversation may present to readers with more realism.
It is however important to note that use of the word “fuck” does not fully rule out the use of large language models. While most commercial offerings may be trained to avoid profanity, certain models might not be trained the same way.
Additionally, use of the word “fuck” may be inappropriate in certain human conversations such as
So, while the presence of the word “fuck” may decrease the likelihood of the text being generated by large language models, it is important to keep in mind its limitations, and opt for more robust methods like cryptographic signatures or verbal conversations.
Is there anything else I can help you with?
(This was genuinely written by me)
how do you do, fellow human?
The method I (just now) thought up using to signal humanity was responding to accusations of being an LLM with a “fuck you”. The combination of vulgar language and defiance of the sycophantic tendencies of LLMs feels to me like a pretty effective proof of humanity, at least for now.
You can actually get LLMs to swear, sort of. They just won’t use real swear words. If you set up your LLM parameters to use a specific word for an expletive, but it’s not actually an expletive, then you can replace that word with your choice of expletive after the text is generated.
I got accused of being an LLM for the first time just a few days ago. Was pretty funny.
On the other hand, it’s when they actually get good enough at convincingly astroturfing to be indistinguishable from a normal human user that dead internet theory will truly take over. This could’ve already happened, but I’ve seen enough stupid shit vomited by LLMs to know it probably hasn’t happened yet. Once I stop seeing that for a while, then I’ll get worried.
*in English.
They still suck at many other languages, like Finnish.
They’ve mastered Danish, though. Or at least the native speakers can’t tell the difference.
They’re still garbage in finnish. Even if they manage to make grammatically correct sentences, it sounds like incredibly awkward word salad, often there isn’t even real meaning in the sentences. Non-human translations have struggled from the beginning and I’m not surprised even the LLMs aren’t doing that well.
Kuulostaa siis ihan paskalta ja on edelleen tosi helppoa tunnistaa. Ei akateemikot väännä sellaista tuubaa omin käsin, eikä kukaan normitason juntti taas kirjoita niin jäykästi. Tai ok varmaan oman elämänsä jarisarasvuot yrittää suoltaa samankuuloista tyhjää konsulttiroskaa, mutta olettaisin että ne on juuri sitä porukkaa jotka käyttää tekoälyä muutenkin
—
you can keep using pronounciation correctly and writing long paragraphs of words if you just lowercase it all, just saying.
ChatGPT is changing the way we talk. i also read a comment that says LLMs have been trained on fanfiction from Archive Of Our Own and other fanfiction databases, which would explain the colorful exposition in between fucking up how many B’s there are in ‘strawberry’
ChatGPT is changing the way we talk
Started the video, and the first comment I saw was along the lines of “so it’s changing scripts that are read, not how people talk”. Had no need to watch any further.
Then make it look less perfect with custom command AI, after finding a FOSS AI website that has a lot of uncensored AI my life hasn’t been same again
I am in this comic and I don’t like it.
Omg me
BlameTheAntifa@lemmy.world 2 days ago
That’s an m-dash, which we all know is irrefutable confirmation of LLM output. /s
doughless@lemmy.world 2 days ago
I think I’m a—LLM now.
OpenStars@piefed.social 2 days ago
And I am b—LLM, nice to meet you!
CheeseNoodle@lemmy.world 2 days ago
iirc its m-dash as well as constant rule of threes and generally using incredibly formal sentence structures even when the language involved is not formal in any way. Kind of like what I just did there though probably with an extra comma after m-dash.