As Iโve already said in a different reply to you elsewhere, something like that already exists (although itโs largely unrelated to the discussion here).
The 
is also considered an insult.
ืืืืืง ืืื ืืชืฉืืื ืืงืืืืช, ืื ืืืื ืืืืืช ืฉืืืืฉื ืืจืขื.
In any case, he violated the forum rules.
Generally true, practically it depends on the context.
That is new to me and I donโt believe there is a consensus about that so far.
Some people understand it as such but the purpose in OSM is just to express โI disagreeโ. It has a long history in OSM and just replaces the comment โ-1โ of the old forum where reactions to post were not possible.
In principle yes but that is definitely not an argument to establish another emoji providing an opportunity for misuse.
Who? The participant posting a reply without using AI or the participant reacting with an bot emoji believing that AI had been used for this post?
I did not talk about intentional misplacing of the reaction here but about the potential for use in misunderstanding and as such causing offence. In such case no one can be blamed of violating the rules but one can still feel offended.
Just as there are many +'s, there should be more of them.
There is a lot of misunderstanding, you canโt limit things because of that.
The same thing is that I disagree because itโs not [this can also be considered a compliment]
Check this out [I think there is broad agreement to remove those who are too lazy to write themselves]
stop threatening me with a good time
14 Likes
7 Likes
The computer is blocking the link.
Strange, it shouldnโt. The only thing I see that would get blocked by the site is ChatGPT.
3 Likes
Regarding my post you mentioned. While I do use AI, I use it almost exclusively for research and evaluation and critique of my tone (Iโm autistic so it functions as an accessibility tool due to my inability to read tone in the text I write and my massive propensity to be very overly direct). I rarely copy directly from the output of an LLM and instead insist on spending a lot of time making my words cognitively accessible and formatting them for the forum so theyโre as easy as possible to parse. Honestly while writing that one I was clearly going a bit mental, my formatting is kinda all over the place compared to my usual and I did not spend as much time reviewing it as I normally would, hence the mistake flagged by a community member later on.
This is often why the conclusions to my posts will read a bit cheasily, itโs really the only section I do struggle with massively and while I usually try to write it myself for this one I had a migraine and did just review it and paste it in a moment of weakness. I think for the purposes of this forum I will be strictly avoiding copying any direct output from now on. Apologies for doing so before.
I havenโt read the full discussion here but we should absolutely be on the lookout for LLM/AI use in forum posts and be flagging it, especially where the body is clearly copied directly from the LLM. LLMs are a bit shit as anything other than a review, critique and summary tool and even then they need to be taken with a grain of salt. They should absolutely not be used for generating an entire proposal.
I am very sorry that original post fell quite so far below my usual standards, honestly reading it back I kinda hate it now. I mainly just wanted to post something to get the ball rolling and kinda spam typed at my keyboard after a v frustrating conversation with Imagico in the OSM Carto repo. I didnโt really do much in the way of proofreading or checking my formatting. The saying โif Iโd had more time I would have written a shorter letterโ comes to mind. I believe I typed out a quick version and then improved each section individually hence it feels a bit disconnected and the formatting is kinda all over the place.
Usually my process for writing a post looks something like:
- Write my immediate reaction down and donโt stop, just generate ideas.
- Take the first draft and turn it into bullet points.
- Sometimes here I will ask an AI to format my work into suggested prose versions based off the original bullet points. These are used for reference only and to help organise the points in my head. I do this less now but itโs one of the parts of writing I really struggle with being dyslexic.
- Generate a set of headers for the discussion to organise my bullet points into.
- Write the final version using the bullet points as a guide.
- Get feedback on tone and the factual content of my work using an AI (no point forcing a community member to check this if an AI can spot it.)
- Iterate, often multiple times, Iโm very particular and often try to be strategic here.
- Format the post to make it as accessible as possible. This is when I add all those bolds, italics, and break up long winding paragraphs into bullet points as well as check that each header is relevant to the text below it.
- Iterate again if Iโm not happy with the final result.
Unfortunately this is what I suspect my process for that original post was like:
- Write like a lunatic taking no structuring into account.
- Format and haphazardly copy paste ideas around all the time lengthening and adding to it using AI feedback and overly generic phrasing far too much.
- Skip my anti-AI proofread. I usually avoid bolding the initial points of a bullet point or writing just 3 even though I have a natural propensity to do so. I may be able to find some of my writing from before AI but I do genuinely write like an AI model when left to my own devices.
- Attempt to write a conclusion and have zero energy left to be positive about the process so it comes out bitter as a radioactive lemon.
- Replace it with an AI generated one and tweak that a little.
- Post with zero proofreading or iteration. (My goal was to get something posted even if it wasnโt perfect to draw attention to the issue.)
Apologies if this post also feels like a bit of a detachment from my normal style, Iโve deliberately not formatted it much or used any of the features mentioned in the initial post to demo my ability to read, comprehend and reply to a conversation without the use of an AI tool and while I think formatting and breaking up concepts is important Iโm aware doing it here would likely be counterproductive if my previous post was perceived as AI. (I just proofread this bit and look at that last sentence, itโs like the LLM is in me. I suspect my years writing and publishing fanfic, which is what lots of AI is trained on may be to blame.)
I would love feedback on this process of engaging with the forum more, I use LLMs in a very specific way to enable me to engage with the community more. However if it is largely the will of the community that AI should not be a part of the process at all I can probably adapt though I cannot promise I will be able to keep my level of engagement the same while doing so.
This was written in a singe attempt without the use of AI/LLM for review or generation purposes or my usual processes. Please be aware my tone may be off in certain sections and I may have repeated myself or otherwise made mistakes I would usually catch. Itโs also much longer than I would usually like for a reply.
EDIT: OMG I hate writing without my normal process, I was trying to resist the urge to proofread this again, but I have and I hate it. Feels v vunerable writing without my normal attention to detail.
3 Likes
Just as a bit of context, the area:highway post linked above didnโt strike me as โmainly AI writtenโ because it actually said something meaningful (unlike many of the other examples linked in the first post at the top of this thread).
1 Like
I want you to know that I also use AI extensively for writing assistance. I think there are two ways that people generally tend to use it: for generating text and generating ideas. I think itโs fine to use either, separately, but just never together.
If you ask an AI โwhat are some reasons that area:highway is a [good/bad] idea?โ and you learn something from that, or it sparks some ideas, great!
If you tell an AI โhere is a forum post Iโm about to write. How does it sound? How could I strengthen my arguments?โ, I think thatโs pretty good too. Or, โhow can I make this sound a bit softer or more compelling?โ OK fine.
The issue comes when you say "please write a forum post in favor of rendering area:highway, youโre gonna get something very generic-AI sounding, unless youโre really really good about editing things to sound in your own words. Thereโs a bunch of speech patterns that are very very common in ChatGPT-generated prose that are fairly unusual in normal human usage.
I applaud you for being forthright about your use of AI. It did come off to me as โarguments written by ChatGPT and then edited by handโ but AI-based nonetheless. The fact that people interacted with it normally (as Andy notes above) does at least tell you that you hit the mark with reasonably cogent arguments.
3 Likes
Bingo! Just as I was reading that I thought โI bet I know someone else who isโ 
2 Likes
As long as you keep having fun with speculation, Iโm all for the joy of it 
I beg to differ. I obviously use LLMs. But more on a Ollama Basis.
Donโt throw me into a pot with OpenAI people.
I use PerplexityAI - and I read through my sources. Sometimes I will make mistakes and misinterpret stuff. Still.
All this thread shows me is, that Milgramโs experiment is still highy popular 
Writing is hard. Not everyone will enjoy reading stuff. Not everyone is up to write a best-seller. Boohoo.
This reminds me of this meme, people who use AI to expand on their few concise thoughts, and other who use AI to break down long texts into few phrases.
Why donโt we just say what we want the way we think, and if the others donโt understand, then try and expand. Donโt try to develop long texts from the very start if what you want to say doesnโt really need that much text to be dealt with. Develop on the go and as you see fit, donโt try to look like a LinkedIn โprofessionalโ (maybe poor choice of example, but thatโs what I thought of atm 
)
(image source: AI Written, AI Read cartoon - Marketoonist | Tom Fishburne )
15 Likes
Exactly. Nice comic btw. Reminds me of the PhD memes
If only we can stop in being procecutors and work together in a true OSS manner - one of my depictions of a greater world.
Iโm all for the second frame. Making stuff easier. Not so biased due to our surroundings.
Iโm definately no stranger to LinkedIn WhatAbout-ism. More often then not I find myself internally screaming โFUTURE BULLSHITโ when someone speaks of smart solutions, AI as a Service and what not. (shout-out to Matthias Horx. Brilliant dude - at least concerning his book(s) )
Growing by the day and getting better at all kinds of stuff.
After all weโre humans and we love to see people fail more then seeing them grow. It is what it is