And yet I honestly engaged with you and even took time to state my point.
And actually, doing so helped me make it clearer! 
Thanks, anyways.
This is an interesting thing to watch and maybe the community might reflect upon its very own behavior. In sociology there is such thing as āgroupthinkā, and āin-group/out-groupā labeling.
We can watch @Mason above posting AI to prove me wrong with (essentially a straw man argument). But interestingly, that āAI slopā got lots of likes quickly, because it seems that Mason is āin-groupā and I am āout-groupā. How dare I ā¦
I disliked the AI language (AI really ruined the word āgenuinelyā and āunpackā for meā¦ though I never really used the latter anyway, heh), but do think there are some interesting philosophical points in there that I could relate to.
I disliked the AI language (AI really ruined the word āgenuinelyā and āunpackā for meā¦ though I never really used the latter anyway, heh), but do think there are some interesting philosophical points in there that I could relate to.
As I like to say, at least one form of intelligence sees the light of day. 
3 is not what Iād describe as a lot of likes, and two of them are from people who also liked your posts, and one of them is from you yourself. I truly donāt get how you came to that conclusion.
But even if I accumulated a lot of likes on that post that you werenāt accumulating, Iād probably just assume itās because people found it to be funny, not that anyone thought it was a serious rebuttal to your point.
But even if I accumulated a lot of likes on that post that you werenāt accumulating, Iād probably just assume itās because people found it to be funny, not that anyone thought it was a serious rebuttal to your point.
I laughed and liked as well.
So maybe, while you are here, you may attempt a serious rebuttal. I am all ears (and eyes)ā¦
Iām not sure I really have a lot more to say on this topic.
I just think that posting unnecessarily long essays is disruptive to conversation, and I think that LLMs supercharge this propensity, and makes engagement with these posts feel even more useless. As Iāve said previously in another thread, I see the guideline about not posting generative AI outputs as really a subset of an existing rule urging people to be concise.
I think that while it often feels like something you arrived at with the help of your LLM is especially clear, the experience of others is often the exact opposite.
I agree, but you introduce a fact out of thin air again āunnecessarily longā, which must not hold ā and does not hold. It depends upon proper use of AI imo.
urging people to be concise
Thatās why I used AI to look over a policy draft so that it might be as concise as possible. My experience with ChatGPT in this regard has been quite good. Not on first try, but if you loop with him and ponder arguments.
I canāt be Business Analyst, Lawyer, MD, Software Engineer, Philosopher at the same time. And adding (!) AI to the loop (AI in the loop) can be extremely helpful imo. I am convinced you can rule about it as much as you like - itās here to stay. I see it as a chance more than a risk.
I would rather read imperfect or even broken English than unnecessarily long AI written comments. They all look the same, they all follow the same template even if you tell it not to. It just becomes a blur for me and I stop reading. Using AI as a tool, fine. Using AI to translate, mostly fine. Using AI for human-human conversation, I think youāre just wasting the other personās time.
The vast majority of AI slop I see is just text inflated to be as long as possible. Seems like a fair statement to me. Itās just unfair to the reader.
You make strong statements and I cannot fully agree. It makes an enormous difference if I make no prompt or if I use ābe a Mario Bunge like AI assistant firmly grounded in scientific realismā¦.ā
If we talk about the legal aspect of say software licenses, then we quickly run into IANAL statements popping up everywhere. Then we continue to discuss what we believe to be our understanding. Right. While AI tools make mistakes, I find them extremely helpful in providing domain aspects that I am not so knowledgable at. Yes, it is not perfect, but I do not consider it a waste of time?
Okay, but what people were telling you repeatedly in that thread was that what you were doing was not increasing the clarity of your communication, nor was it improving the sharpness of your point.
If nobody can tell whether or not youāre writing with the assistance of an LLM, then nobody will care. What people objected to was you pasting text that they felt was very āLLM-eyā, and was detracting from the conversation.
So my policy proposal was a distraction and missing the point? I was not being helpful in raising the issue? I did not try to come to a closure?
Really? Is that what you are saying?
The end of the game is that everything continues as before: Everybody knows that a single LICENSE text cannot meet modern demands and because just checking this is so convenient, weāll leave it at that.
Please, be quiet, man, you used AI to help out. 
I think what people were telling you was that at least parts of your communication were felt to be made significantly less clear by your usage of LLMs. They were not saying that what you contributed was without value, they were just asking you to communicate in a more human manner.
I understand that can feel rude, and that as a non-native speaker it can even be an extra burden on you, but I think what youāre not acknowledging is the burden that this style of communication also places on the people reading it.
Ok, so which prompt did you use here? Clarify RegistryCI license-check API for REUSE/SPDX-compliant packages Ā· Issue #155557 Ā· JuliaRegistries/General Ā· GitHub
Because this is the kind of comment that I am talking about. Worst of all is this sentence:
That is not a useful substitute for a clear author-side statement about the licensing terms under which the package source code is offered as a package.
As a non-native English speaker myself, I cannot even comprehend what this sentence is saying. So I canāt imagine you can?
Yes, I was also accused that Claude invented āOPLEā but it was me. What I mean to convey is that there is no mechanistic rule that can likely tell you what the āoutboundā license has to be, because we do not have a clear way to solve conjunctions, e.g. MIT and GPL-3.0-only and CC0-1.0 means what?
What that tries to say is that registry does not need to check that. It can and should trust what the submitter/author claims (up to some checking that is). The Package can mean āeverything contained in a tarballā and package source code is maybe different, e.g., what you get by āusing Packageā.
It it rather difficult to talk about licensing, as there are different layers to it. Thatās indeed a reason for me to discuss my ideas back and forth with AI and also because in legal discussions not getting terminology right makes the exchange almost nonsensicalā¦
I should add that I am also still new to Julia so that I also have to try to understand the technicalities in the discussionā¦ (I really try to do this).
Right, and thatās precisely why weāre talking past eachother. Youāre simultaneously having a conversation with some AI, and then not actually listening to me. So you get the response from Claude to a different thing than what Iām talking about, and keep going off in a completely different direction without taking time to actually understand what Iām saying. And then you very obviously plopped a very-lightly edited sloppy response right back at me. Thatās frustrating.
That post was purportedly responding to me, but itās completely missing what Iām saying. Iāve been pointing out throughout that thread that we donāt have a way for the submitter/author to unambiguously claim anything to the registry beyond the status quo convention.