LinkedinLunatics

4456 readers

530 users here now

A place to post ridiculous posts from linkedIn.com

(Full transparency.. a mod for this sub happens to work there.. but that doesn't influence his moderation or laughter at a lot of posts.)

founded 2 years ago

MODERATORS

csm10495@sh.itjust.works

crmsnbleyd@sopuli.xyz

381

What is the benefit to spamming posts with LLM generated comments (sopuli.xyz)

submitted 1 day ago by crmsnbleyd@sopuli.xyz to c/linkedinlunatics@sh.itjust.works

35 comments fedilink hide all child comments

you are viewing a single comment's thread
view the rest of the comments

[–] Sgt_choke_n_stroke@lemmy.world 16 points 1 day ago (4 children)

I just think it's odd how many verbs chat gpt uses like "crucial", "essential", and "leverage". Like I don't use that shit in regular conversations or papers. It's like a small hint that it wants to be caught.

[–] itslilith@lemmy.blahaj.zone 8 points 20 hours ago (2 children)

🤓☝️ akshually, crucial and essential are adjectives

[–] sugar_in_your_tea@sh.itjust.works 7 points 19 hours ago* (last edited 19 hours ago) (2 children)

And leverage is a noun. OP needs to back to grammar school.

Also a verb, to be fair. 😅

[–] Sgt_choke_n_stroke@lemmy.world 4 points 16 hours ago

I think my ignorance proves that I'm human ¯\_(ツ)_/¯

[–] Revan343@lemmy.ca 3 points 16 hours ago

Leverage can be a verb, but isn't here

[–] IncogCyberspaceUser@lemmy.world 3 points 19 hours ago

Thank you. I just woke up and was questioning my English skills and my entire reality.

[–] brucethemoose@lemmy.world 12 points 22 hours ago* (last edited 22 hours ago)

LLMs, in fact, have slop profiles (aka overused tokens/phrases) common to the family/company, often from “inbreeding” by training on their own output.

Sometimes you can tell if new model “stole” output from another company this way. For instance, Deepseek R1 is suspiciously similar to Google Gemini, heh.

This longform writing benchmark tries to test/measure this (click the I on each model for infographics):

https://eqbench.com/creative_writing_longform.html

As well as some some disparate attempts on GitHub (actually all from the eqbench dev): https://github.com/sam-paech/slop-forensics

https://github.com/sam-paech/antislop-vllm

[–] phdepressed@sh.itjust.works 14 points 1 day ago

The training data probably includes a lot more formal writing. As the major selling point of chatgpt is it sounding like it "knows" things. More "complex" verbiage is helpful to that. This type of writing is more common in things like textbooks and scientific writing in general which have been at least part of its training data.

[–] sugar_in_your_tea@sh.itjust.works 2 points 19 hours ago

Yeah, it's overly formal, but I do use each of those in regular conversion, just a lot more sparingly than AI seems to.