Reminder: #LibreOffice includes no AI features in the software.
-
Reminder: #LibreOffice includes no AI features in the software. It doesn't rewrite your content, train algorithms, or share your data. There is zero AI functionality in the app.
Regarding the use of LLMs in the development process, we have a strict policy about this: https://wiki.documentfoundation.org/Development/AI_policy – Contributors who use such tools must be able to explain EVERY changed line of code. We do not accept any "slop".
@libreoffice
stop redefining our feedback to suit your narrative. we want NONE of this. your policy should be a flat "this is not allowed", PERIOD. no alternatives.instead of trying to do damage control PR as if you're trying to sell me something, you should just fix the damn problem. that's all we care about.
-
Reminder: #LibreOffice includes no AI features in the software. It doesn't rewrite your content, train algorithms, or share your data. There is zero AI functionality in the app.
Regarding the use of LLMs in the development process, we have a strict policy about this: https://wiki.documentfoundation.org/Development/AI_policy – Contributors who use such tools must be able to explain EVERY changed line of code. We do not accept any "slop".
@libreoffice as Yoda might say: have it both ways you cannot.
-
Reminder: #LibreOffice includes no AI features in the software. It doesn't rewrite your content, train algorithms, or share your data. There is zero AI functionality in the app.
Regarding the use of LLMs in the development process, we have a strict policy about this: https://wiki.documentfoundation.org/Development/AI_policy – Contributors who use such tools must be able to explain EVERY changed line of code. We do not accept any "slop".
It seems like you guys were trying to go for a reasonable compromise, but I really don't think one exists. You allow slop or you don't. You've picked slop*.
I would like to strongly urge you to reconsider this. Your software is important and matters too much for this.
* Which is to say, plagiarism and the potential loss of copyright of your project across several jurisdictions - things which have nothing to do with code quality. You are putting your use of an open license at risk in order to seem "reasonable".
-
Reminder: #LibreOffice includes no AI features in the software. It doesn't rewrite your content, train algorithms, or share your data. There is zero AI functionality in the app.
Regarding the use of LLMs in the development process, we have a strict policy about this: https://wiki.documentfoundation.org/Development/AI_policy – Contributors who use such tools must be able to explain EVERY changed line of code. We do not accept any "slop".
@libreoffice you know a technology is good and cool when you have to write a ten point note on how it must not be used. that's when you know it is ready for prime time.
-
Reminder: #LibreOffice includes no AI features in the software. It doesn't rewrite your content, train algorithms, or share your data. There is zero AI functionality in the app.
Regarding the use of LLMs in the development process, we have a strict policy about this: https://wiki.documentfoundation.org/Development/AI_policy – Contributors who use such tools must be able to explain EVERY changed line of code. We do not accept any "slop".
@libreoffice That's really sad to read. What's the point of FOSS if we're going to be dependent on LLMs (i.e. the worst of Big Tech) for any future development anyway? Because yes, that's unavoidable once you open the gates to LLM-written code. Human contributors *will* start to lose control and knowledge of the codebase.
-
@libreoffice "While the use of generative AI based on Large Language Models in software development has several issues, they are allowed in LibreOffice contributions"
I'm sorry but I'm afraid the "with certain restrictions" which follows does not make it not-slop. It is what it is u.u
@VileLasagna @libreoffice Ah, so a if AI writes 1+1=2, it's not true anymore, and jUsT sLoP? You guys are completely losing track ...
-
Reminder: #LibreOffice includes no AI features in the software. It doesn't rewrite your content, train algorithms, or share your data. There is zero AI functionality in the app.
Regarding the use of LLMs in the development process, we have a strict policy about this: https://wiki.documentfoundation.org/Development/AI_policy – Contributors who use such tools must be able to explain EVERY changed line of code. We do not accept any "slop".
@libreoffice Still slop, guys. Still sending training data to who? Claude? Codex?
Still fucking up the world.
-
@davidgerard @DamonWakes @libreoffice Ah, so a if AI writes 1+1=2, it's not true anymore, and jUsT sLoP? You guys are completely losing track ...
-
Reminder: #LibreOffice includes no AI features in the software. It doesn't rewrite your content, train algorithms, or share your data. There is zero AI functionality in the app.
Regarding the use of LLMs in the development process, we have a strict policy about this: https://wiki.documentfoundation.org/Development/AI_policy – Contributors who use such tools must be able to explain EVERY changed line of code. We do not accept any "slop".
@libreoffice LibreSlop.
-
Reminder: #LibreOffice includes no AI features in the software. It doesn't rewrite your content, train algorithms, or share your data. There is zero AI functionality in the app.
Regarding the use of LLMs in the development process, we have a strict policy about this: https://wiki.documentfoundation.org/Development/AI_policy – Contributors who use such tools must be able to explain EVERY changed line of code. We do not accept any "slop".
@libreoffice Started using LibraOffice about 3-4 years ago to reduce annoyances from MS Word. And loving it! Also grateful that it's compatible with Word so stuff I write can be read by people I work with.
-
@rzeta0 @libreoffice You’re being sarcastic, right?
-
@davidgerard @DamonWakes @libreoffice Ah, so a if AI writes 1+1=2, it's not true anymore, and jUsT sLoP? You guys are completely losing track ...
@kingsgambit @davidgerard @DamonWakes @libreoffice do you need an llm to write 1+1=2 ?
-
@VileLasagna @libreoffice Ah, so a if AI writes 1+1=2, it's not true anymore, and jUsT sLoP? You guys are completely losing track ...
@kingsgambit Yes, very astute of you! If you read my toot, you'll find out that this is exactly what I meant and this response is not at all an attempt at a distraction by derailing the conversation into arguing irrelevant semantics!
Well spotted, I stand defeated forever and, as penitence will allow OpenAI to install ChatGPT in the brain of my firstborn child..
-
@kingsgambit @davidgerard @DamonWakes @libreoffice do you need an llm to write 1+1=2 ?
@peach @davidgerard @DamonWakes @libreoffice you get the point. If a human has put in the work to understand and verify each line of code, but maybe the implementation was done in an elegant way by an LLM in short time, then that is good code. The product is what counts. The fear behind SLOP callers is just that the code is unexplainable and mysterious, and you fix that by human review.
-
@kingsgambit Yes, very astute of you! If you read my toot, you'll find out that this is exactly what I meant and this response is not at all an attempt at a distraction by derailing the conversation into arguing irrelevant semantics!
Well spotted, I stand defeated forever and, as penitence will allow OpenAI to install ChatGPT in the brain of my firstborn child..
@VileLasagna you implied that it's still slop though, even with their restrictions, which require human review and understanding. So that fully translates to my example.
-
@peach @davidgerard @DamonWakes @libreoffice you get the point. If a human has put in the work to understand and verify each line of code, but maybe the implementation was done in an elegant way by an LLM in short time, then that is good code. The product is what counts. The fear behind SLOP callers is just that the code is unexplainable and mysterious, and you fix that by human review.
@kingsgambit @davidgerard @DamonWakes @libreoffice the term SLOP is the process though. It’s the lack of care and just using models that are 100% based on stolen works. The ends don’t justify the means. Sorry I don’t make the rules.
-
@VileLasagna you implied that it's still slop though, even with their restrictions, which require human review and understanding. So that fully translates to my example.
@kingsgambit Oh, apologies for the confusion, let me clarify:
Yes, it IS slop.
Whether the slop appears functional or correct of truthful or whatever is irrelevant. Slop is slop, that's just what it is.
Libreoffice is there arguing that they only allow the best slop in, you see, so it's fine. I am not interested in how much they like the slop that they approve, I still find said slop deeply repulsive.
Definitionally, LLM output is slop. People's opinions on the content don't change that fact
-
Reminder: #LibreOffice includes no AI features in the software. It doesn't rewrite your content, train algorithms, or share your data. There is zero AI functionality in the app.
Regarding the use of LLMs in the development process, we have a strict policy about this: https://wiki.documentfoundation.org/Development/AI_policy – Contributors who use such tools must be able to explain EVERY changed line of code. We do not accept any "slop".
"If you intend to use an LLM to make lots of patches in one area, start small to avoid having to make many changes after code review feedback."
As in you expect to have to change it more than usual code because it's so ass??
-
@kingsgambit @davidgerard @DamonWakes @libreoffice the term SLOP is the process though. It’s the lack of care and just using models that are 100% based on stolen works. The ends don’t justify the means. Sorry I don’t make the rules.
@peach @davidgerard @DamonWakes @libreoffice That's fine, but it's like deciding to keep running in a race where others are using cars. Sure, they might not be very talented, driving is not that complicated ("lack of care"), but they will overtake you. And leave you with a vulnerable, left-behind software product, if we shall leave the analogy and get back to the actual application here.
-
@kingsgambit Oh, apologies for the confusion, let me clarify:
Yes, it IS slop.
Whether the slop appears functional or correct of truthful or whatever is irrelevant. Slop is slop, that's just what it is.
Libreoffice is there arguing that they only allow the best slop in, you see, so it's fine. I am not interested in how much they like the slop that they approve, I still find said slop deeply repulsive.
Definitionally, LLM output is slop. People's opinions on the content don't change that fact
@VileLasagna Okay but then, since you are not arguing on the quality of the code and rather on the semantics of who wrote it, it's a little pointless as a critique? That's like saying "the code might be good but my neighbour, who I hate, wrote it, so it's shit".
