Bryan Davis1d ago
Baldur Bjarnason

One of the lessons that many in software strangely never learn is that breaking something while being an asshole gets a very different response from people than breaking something while not being an asshole

And, given how “AI” continues to pour shit on the rest of society, using LLMs is widely going to be perceived as an asshole move

For better or for worse, using LLMs in OSS is going to be seen as an adversarial move that undermines community building

Jun 7 at 9:59amWeb
Show thread𝓙. 𝓜.1d ago
@baldur the way I have seen it put is that "if you couldn't be arsed to spend the time writing the code yourself, why should we bother to spend the time reviewing it?"
Show threadLesley M 🏴󠁧󠁢󠁳󠁣󠁴󠁿 🇪🇺1d ago
@baldur Some people haven't connected the dots re A.I. . The mass rush for A.I. datacenters is making the PC hobby completely unaffordable, with RAM, Storage, and GPU prices becoming unaffordable for the majority, and building resentment. I think a lot of peeps see using A.I. as supporting the very thing that is hurting everyones pockets right now, especially in some US states where Electricity cost are through the roof for infrastructure upgrades. ones that the A.I. companies should pay for.
Show threadTheriac1d ago
@baldur
#technicallyOC


ALT for more context/info.



#AI no ai was used in the creation of this post though I presume anything attached to this account has already been trawled and I hope that by tagging it as AI, in my own small way, I manage to poison someone's LLM.
Show threadjtb1d ago
@baldur I disagree. Much time is wasted making syntax errors. AI doesn't make syntax errors. I expect there was similar resentment when high level languages took over from assembler.
Show threadchx1d ago
@jtb LLMs do not make syntax errors? *in your dreams*
Show threadjtb1d ago
@chx Not the one I was using. It wrote the code and compiled the code and ran the code and tested the code and edited the code and reran the code.
Show threadLeeloo1d ago

@chx @jtb
To be fair, it probably does make fewer syntax errors than someone so new to development that they think syntax errors is where any relevant amount of time is being spent.

That's pretty much on par with claiming that self driving cars are great, because you don't have to waste time trying to remember which pedal is the brake and which is the throttle every time you come up to a stop sign.

Show threadjtb1d ago
@leeloo @chx Perhaps you only write programs in one language. Try writing programs in a language you aren't familiar with. I know it is a waste of time arguing though, you are AI haters, you think in black and white, for you, it's a good vs evil thing.
Show threadchx1d ago
@jtb @leeloo it is indeed evil. You got it in one. And I will wear the LLM hater t-shirt with pride. I refuse the AI hater marker: a lot of success in my life comes from Stanford NLP. It was and still is a useful toolset which never instructed a child to commit suicide, exploited Africans or poor communities for data centers. Funny how that works and funny how LLM proponents want to piggyback on that success by equating LLMs with AI.
Show threadLeeloo1d ago

@jtb @chx
Only one language? You mean other than the four I have listed on my profile as my favorite languages?

You have no clue what you are talking about.

Show threadchx1d ago
@jtb @leeloo as for a single language, I learned Pascal, my second programming language on a 8 bit home computer when I was 11 years old and back then I was still living behind the Iron Curtain. Yes. It was that long ago. So perhaps no, I am not writing programs in a single language.
Show threadJohn Maxwell1d ago
@jtb @baldur Horseshit.
Show threadjtb1d ago
@jmax @baldur bollox
Show threadfancysandwiches1d ago
@jtb @baldur if you're wasting time on syntax errors in 2026, with, or without LLMs, then you are doing something woefully wrong. All modern IDEs and Editors have had good enough auto-complete that syntax errors are essentially a solved problem, and have been for years.
Show threadAssimilateborg1d ago
@baldur each time I suggest to use LLM to resolve a problem, coworkers more and more react with negativity. I love that! We are slowly going back to normality.
Show threadmirabilos🐈‍⬛1d ago

@assimilateborg @baldur dann schlag’s halt net vor.

Wobei… wenn ich mir das so ansehe…

[… I]n my long quest to find AI any use, I found "one shot usage programs". So you generate them once. […]

… dann bist Du auch ein LLM-Promoter.

Show threadAssimilateborg11h ago
@mirabilos @baldur und du fährst nur mit der bahn weil alle anderen "auto-promoter" sind wenn die es doch manchmal nützlich finden mit dem Auto wohin zu fahren wo die bahn nicht hin kommt.
Jeder Scheiss hat irgendwo doch eine Verwendung.
Und ich selbst nutze nicht mal diese one-shot LLM Sache weil das nicht mein Job ist. Hab nur die Konsequenzen davon zu spüren bekommen (viele Updates kurz nacheinander)
Show threadKoen Hufkens, PhD1d ago

@baldur But they will have communities of agents... cotcha  

*sarcasm*

Show thread707Kat1d ago
@baldur especially when the system they utilise is so fragile that plain english is looked on as “malware” and they can’t even be arsed to read the readme of the github repo before having their agents fetch.
Show threadMx. Aria Stewart1d ago
@baldur Sounds like a good adage here is "don't break things, spend the time getting it RIGHT"