idk about being a straw-man, but regardless the reply was addressing the misleading and not giving proper credit to the researchers and further giving that LLMs were used for analysis, not full on finding the exploit, so no LLMs aren't good at finding exploits without clear search inquires by humans.
As for the empathy and the robo-sexuality- It was the intentional point of the original comment that people find heavy social relation towards LLMs or other objects that are able to communicate back to them. Even in our examples of the movies they touch on romantically/sexual relations towards robots and a couple others point towards the empathy of them as well. PS these are topics from 1950's not "whatever the shit kids are in to these days." Most people affected by this are older generations and young adults without social netting.
Turning it around phrasing that LLMs are useful towards finding exploits makes it sound more like your wanting to use LLMs for using said exploits rather than using LLMs for better use cases. Regardless its still not possible nor ever will be because again LLMs can only use predetermined variables based on its previous learning data set and random variables (PS those random variables that are undesirable are what is commonly called hallucination, its just unwanted variables in a huge spaghetti code.) Its even on the site your sourced:
"Was this AI-found? AI-assisted. The starting insight — that splice() hands page-cache pages into the crypto subsystem and that scatterlist page provenance might be an under-explored bug class — came from human research by Taeyang Lee."
If we misread your interpretation then our mistake, however the phrasing felt more that you were praising AI for finding exploits and not for actual good use and it read out to us like an ethical issue.
If making this stance clear that LLMs make more harm than good in the case of chat Bots and being used as full on replacements of people makes us a Straw-man than IG we're a straw-man or whatever lol.
Though we can probably agree that Machine Learning can, should and have been used since the 1950s as glorified search and calculation engines for complex equations and datasets. They can make really good use for generating and categorizing random protein molecules, find patterns in cancer research and even filter out examples astronomers find in the night sky; however its overall useless without a qualified and passionate researcher who knows their stuff and can double check their ML sifters.
Sources for the saucy beans:
- 2016 BBC report on rise of Romantic Robotics: https://www.bbc.com/news/uk-england-berkshire-35354263 (wild fucking article, the movies give more light to the subject than the article, but we included it as its an interesting read)
- Your own source for ease of reference:https://copy.fail/#faq (second to last drop down in FAQ titled: Was this AI-found?)
- 2020 History of Artificial Intelligence in Medicine https://www.giejournal.org/article/S0016-5107(20)34466-7/fulltext (Really good read on the history of Machine Learning in medicine)
- 20205 AI advance helps astronomers spot cosmic events with just a handful of examples https://phys.org/news/2025-10-ai-advance-astronomers-cosmic-events.html (AI helped filter and search for patterns it was taught from prior examples and its output was still look through and refined by proper astronomers)
^edit, fixed a bit of formatting lol^