Comment on An earnest question about the AI/LLM hate
Fontasia@feddit.nl 2 weeks ago
I know there’s people who could articulate it better than I can, but my logic goes like this:
- Loss of critical thinking skill: This doesn’t just go for someone working on a software project that they don’t really care about. Lots of coders start in their bedroom with notepad and some curiosity. If copilot interrupts you with mediocre but working code, you never get the chance to learn ways of solving a problem for yourself.
- Style: code spat out by AI is a very specific style, and no amount of prompt modifiers with come up with the type of someone really going for speed or low memory usage that’s nearly impossible to read but solves for a very specific case.
- If everyone is a coder, no one is a coder: Of everyone can claim to be a coder on paper, it will be harder to find good good coders. Sure, you can make every applicant do FizzBuzz or a basic sort, but that does not give a good opportunity to show you can actually solve a problem. It will discourage people from becoming coders in the first place. A lot of companies can actually get by with vibe coders (at least for a while) and that dries up the market of the sort of junior positions that people need to get better and promoted to better positions.
- When the code breaks, it takes a lot longer to understand and rectify when you don’t know how any of it works. When you don’t even bother designing or completing a test plan because Cursor developed a plan, which all came back green, pushed it during a convenient downtime and has archived all the old versions in its own internal logical structure that can’t be easily undone.
mbtrhcs@feddit.org 2 weeks ago
I’m an empirical researcher in software engineering and all of the points you’re making are being supported by recent papers on SE and/or education. We are also seeing a strong shift in behavior of our students and a lack of ability to explain or justify their “own” work