@darth Your work would benefit greatly from a clear declaration of whether or not AI assistance was used. With this writing style, I can't help but slopvestigate

@ariarhythmic I used to mark it when I used LLM's and then suddenly the entire Internet became A LOT bigger slop than my blog will ever be so I thought to myself — why bother?

But yes, Claude is involved.

Now that we got that sorted out please tell me, as a blog reader, how much does it bother you knowing that I am not a native English speaker and that I have decided to swap LibreOffice dictionary for something like Claude? I need to understand this from your perspective so I can adjust.

@ariarhythmic didn't fit 500 characters, but I generally want to add value to Free Software community. I teach people Linux every Sunday live on stream for 5 hours straight. I make howtos that eat the majority of my free time and on top of that I make blog posts that are, in essence, a summary of what I said in my video.

So knowing all that — please tell me whether you would be more likely to read something I wrote with zero assistance or how much does this matter?

@darth @ariarhythmic I don't want to speak for the OP, but their request was pretty straightforward: if an LLM is helping with the writing, disclose that.

If you don't want to do that, fine, it's your blog. (It makes me want to not read it, but still.)

But if that's your position, just own it. Don't fall back on excuses.

@ariarhythmic I am saying that these are my sentences. You can hear them in the video. But Claude altered them.

@darth did you use an LLM to edit parts of this? I get a little bit of the LLM smell but they *were* trained on real humans and there's not enough for me to definitively think you did or didn't

@jalefkowit @ariarhythmic if it's important I will disclose it. My reply isn't an excuse but an attempt at having a conversation. I wish to find out what is important to people.

And how do I disclose what I did? Which category is it if the content is mine but grammar isn't?

@Byte I am having this same parallel conversation with another person. These is Claude in there, yes. I don't know if you watched the video, but the process was that I said what I said in the video — from my own mind as I usually do. Then Claude cleaned up the transcript.

I just learned that people want any % of Ai disclosed. I will do that when I get home.

@darth I don't want to read or watch anything that had any amount of AI involved in it's creation.

I'm disappointed to hear you debased yourself and your writing in this way.

Hire a human editor, or do it yourself, but for fuck's sake, don't reach for an LLM.

@darth @ariarhythmic you can't "add value to [the] Free Software community" using a tool designed to exploit and extract value for large corporations. This is not giving anything back, it's just enriching the thieves and being lazy.

@darth @ariarhythmic If you needed assistance from a human co-writer, editor or translator, how would you disclose that? You'd put their name on it somewhere.

It's less important if it's a one-off thing, but if you are leaning on the LLM to produce much or most of your work it seems fair to mention that. Particularly since you are going to get complaints that your text reads like all the other LLM-generated text out there anyway. At least then nobody feels tricked.

@jalefkowit @ariarhythmic true. But maybe this case deserves a little more light and then you can be the judge. Please find at least one minute of time to check my video, it's linked on the top of that blog post. In there you will see me talking about this topic but while I talk it's not "good enough" to be literally transcribed into a blog post. A blog post is then "fixed" from the source and the source is me. Me talking. Me reading names of people from my notes. Perhaps I am guilty. IDK.

@Byte @ariarhythmic I meant by having a classroom — that's my gift to the community. But I know what you meant.

@ariarhythmic @darth that is helpful, yes, but it’s cheapened by your use of LLMs.

There’s no ethical way to use LLMs in the way that they exist currently.

@Byte @ariarhythmic OK, I see that you are willing to teach me something so I will take you up on this offer.

MY videos are the source material that I create. For example I recorded a new video 2 hours ago and it will be live on PeerTube/YouTube later today. Once it's out I will make a blog post about it. I have a couple of choices:

1. use the transcription of my voice and rewrite it as written word
2. give transcription to LLM to clean it (that's what I did)
3. write from scratch, no spellch.

@lolitechengineer @ariarhythmic @Byte that's actually a bigger threat to humanity than you presented it.

@darth @ariarhythmic adding my voice that I did want to read your post after seeing this quoot but now that I know that an LLM was involved I won't. I don't mind if your english isn't that good.

@darth @ariarhythmic

you’re forgetting the option of using a traditional/non-LLM spellcheck program. Like the one in LibreOffice.

There are also some decent grammar suggestions in there usually. But speaking as both a native English speaker and someone who hates AI on principle, I would far prefer bad grammar in English than anything that’s been touched by an LLM.

@ariarhythmic @darth

Ignore them, they’re not worth your time. And they’re saying some pretty fascist stuff