Post #1973012
2026-05-03 18:16 UTC
Replies (4)
-
@blinry@chaos.social 2026-05-03 18:18
Relatedly, I'd like to retain a deep understanding the project and its code base, and get to practice skills like debugging, technical writing, and using the involved languages and tools. This is the fun part for me! :) I've used LLMs for generating code completions in the past, and it has led to me relying on them too much. I want to get more skillful in what I do, not less. Even if it can be exhausting at times.
-
@fiveop@mastodon.fiveop.de 2026-05-03 19:02
@blinry Stochastics was my least favourite subject during my math studies, because I was there to know things precisely (personal preference). And I feel similar about developing software. I want to dot the Is and cross the Ts. Using LLMs feels like saying “80% is good enough; onto the next problem”.
-
@movq@tilde.zone 2026-05-03 19:08
@blinry This is one of the main reasons why I don’t use AI tools: It’s very, very boring. No, I want to dig through the problems and I want to solve them. Having some magic machine do that for me makes the whole endavour completely pointless. (And then there’s all the other reasons on top …) I am somewhat optimistic that more people will realize this as well over time. How dull their hobby has become. Or maybe they won’t, because only a certain kind of personality actually cares about this, I don’t know. 🤷♀️
-
@jpl@norden.social 2026-05-03 19:32
@blinry How does this argument not exclude *any* contributions by someone else than you? Because if I wrote something by hand (which is the only way I write anything), you wouldn't have worked through the problem either - I would've. Is it important that I am proud of the result, too? It's a good argument for you not using LLMs yourself (and probably one of the reasons I'm not using them), but I don't see how this applies to contributions by others.