LLMs are going to make senior devs indespensable. So far from what I’ve seen, it’s not great at solving unusual cases, and it most shines in boilerplate and generic problems.
So juniors are never going to learn to code, and then companies will have to pay for experienced people.
Juniors never think hard about unionizing, and the seniors will have job security and therefore not strong motivation.
I hope devs will unionize in any case, LLMs or not, like any other specialization.
jsomae@lemmy.ml 4 days ago
I think it’s because most programmers use and appreciate the tool. This might change once programmers start to blame gen AI for not having a job anymore.
mke@programming.dev 4 days ago
There remains a significant enclave that rejects it, but it’s definitely smaller than equivalent groups in other mentioned professions. Hopefully things won’t get that far. I think the tech is amazing, but it’s an immense shame that so many of my/our peers don’t give a flying fuck about ethics.
nickwitha_k@lemmy.sdf.org 4 days ago
Reporting in.
Yup. Very much agreed here. There are some uses that are acceptable but it’s a but hard to say that any are ethical due to the ethically bankrupt foundations of its training data.
takeda@lemm.ee 4 days ago
I noticed a bad trend with my colleagues who use copilot, chatgpt etc. They not only use it to write code, but also trust it with generally poor design decisions.
Another thing is that those people also hate working on existing code, claiming it is communicated and offering to write their (which also ends up complicated) version of it. I suspect it’s because copilot doesn’t help as much when code is more mature.
dev_null@lemmy.ml 4 days ago
And programmers retain complete control of the output - it’s just a bit of text that you can adapt as needed. Same as looking up snippets from Stack Overflow. Programmers are used to finding some snippet, checking if it actually works, and then adapting it to the rest of their code, so if doesn’t feel like introducing media that you didn’t create, but like a faster version if what everyone was already doing.