Mastodawn

Arseny Kapoulkine

I don’t fully understand the motivation to use LLMs to write blog posts. The loss of personal voice is so striking that it immediately rings alarm bells in my head. You can fact check the output although even that is spotty but you can’t inject the soul back.

And if you stop writing yourself, what makes you think you still can? And what is it that makes you - you?

Daniel Gibson Feb 18

@zeux
Even more fundamental: Why "write" a blogpost if you don't want to write a blogpost?

Arseny Kapoulkine Feb 18

@Doomed_Daniel At this point I would literally take photos of napkins with jotted notes over whatever the LLM produces when you feed the photos in.

Daniel Gibson Feb 18

@zeux if there even were any notes with any thoughts and not just "write a blogpost about XYZ that I can share on LinkedIn"

Alexandre Mutel Feb 18

@zeux I haven't used LLMs for my blog posts yet. But I might use one as a helper for my English. As a non-native speaker, it could help me rephrase sentences, fix awkward wording, and... make things clearer in the end.

The content and the point of the post would still be mine. I wouldn't use it because I don't know what to say or where to start - only to express what I already mean in better English.

Alexandre Mutel Feb 18

@zeux And for the "soul" part: I think you can guide it a bit. For example, I could give it a couple of my previous posts and ask it to keep a similar tone and style. I'm pretty sure that could work reasonably well… but I'll have to try. 🙂

@xoofx @zeux I'd much prefer reading blog posts in "worse" English tbh, that would at least immediately signal a human wrote it

@tracefree @xoofx @zeux same, mildly broken English is fine.

Arseny Kapoulkine Feb 18

@xoofx Using LLMs to fix up grammar is fine, sure. For everything else, I suggest trying first and coming up with your own criteria; my perspective is simple. It's human to err, and we used to think improving your writing is part of the process. Your "awkward wording" is just fine actually.

Neil Henning Feb 18

@xoofx @zeux as someone who barely speaks what the English call English - I prefer to read stuff that is authentically you even if it’s not proper according to the linguists.

Ignacio Castaño🍉Feb 19

@neilhenning @xoofx @zeux I feel like this is targeted at me. 😅

My problem is that I'm not comfortable releasing something that I know has mistakes, so I usually re-read and rewrite my posts a hundred times until I'm satisfied with the wording, and this takes a long time.

Being able to have an LLM correct all my typos and awkward sentences is a huge time saver. I don't see it as a choice between broken English and LLM impersonation, but between minutes and hours.

Ignacio Castaño🍉Feb 19

@neilhenning @xoofx @zeux Do my posts loose some of its personality? Maybe. I do not accept all the suggestions blindly, but only the ones that feel right, or that I feel I would use myself. So, I hope some of it is preserved.

Josh Simmons Feb 19

@neilhenning @xoofx @zeux Not to mention it's not actually good at fixing those errors... It just introduces a whole new class of mistakes. Just because the computer generated it doesn't mean it's right.

Alexandre Mutel Feb 19

@dotstdy @neilhenning @zeux well, it is definitely much better than what I could write, and I'm re-reading it, and I can tweak it to my taste. It can save hours frankly, and my spare time is thankful for such help. ☺️

Josh Simmons Feb 19

@xoofx @neilhenning @zeux why do you write? how can you save hours doing something more or less completely optional :')

Josh Simmons Feb 19

@xoofx @neilhenning @zeux anyway, it's whatever, but do keep in mind that at least for me, if I notice that somebody has used an LLM to generate a piece of writing (and it's usually very obvious), i just close the tab. So that saves me a lot of time too, in that sense I guess we're all saving time.

Alexandre Mutel Feb 19

@dotstdy @neilhenning @zeux I do the same. I'm not talking about using LLM to generate shit, unless this is your vision of using LLM in general. Anyway, I think, in general on Mastodon, we cannot talk about LLM because it is a sensitive topic, so I will avoid that topic entirely.

Neil Henning Feb 19

@xoofx @dotstdy @zeux you can totally talk about it to me I use them all the time (but I get there is a good cohort of people that really hate LLMs and find no value in them which is mad to me!)

Alexandre Mutel Feb 19

@dotstdy @neilhenning @zeux good question, many things are optional in our life. I believe I wrote (not much anymore) to share projects and talk about inner details about what I discovered during these projects. I know for sure that making these blog posts always took me several hours of work (over multiple days).

@zeux
I guess it really depends how you use it.
I usually write the post and ask the LLM to review, make sure it's clear and format.
In my last post, I wrote 300 pages and asked to summarise as a post. Later I did some adjustments and it's done.
Otherwise, I would probably not find time to post anytime soon and, honestly, most times, AI quickly improves the result. Often, AI rewrites parts better than my original writings.

Arseny Kapoulkine Feb 18

@ygamedev I can't address this without commenting re: your actual post so I won't. Other than to say: it's completely fine to post nothing if you don't have time for it.

Matteo Tuzzato Feb 18

@zeux there's a monster in this forest ...

Anton Palmqvist Feb 18

@zeux As someone wise once said: ”Why would I bother reading something no one bothered writing”

Stephen Brooks 🦆Feb 19

@zeux It's like "tell me you have nothing to say without telling me you have nothing to say".

André-LA 🎀 gamedev Feb 19

@zeux The ones I found were clearly just an attempt to be a "influential" voice, which mostly become concentrated on these "water is wet" LinkedIn articles.

@zeux
same, I don't even like to proofread mine with llm tools or something. *I* have to learn if there is a mistake.

NeonkAaa Feb 19

@zeux One may say "you could train the LLM to mimic you, so you can focus on storytelling"