Oh it's very much that. Every single study that's come out on companies adopting AI shows that it makes no meaningful difference to productivity. So, it's very clearly just an excuse to do layoffs.

[–] hexaflexagonbear@hexbear.net 4 points 20 hours ago (1 children)

Coding tools seem impressive, but it’s still very much only useful to get a rapid prototype held together by duct tape. The other side cases are low-key hilarious. We’re pressured to meet vendors constantly, and you can tell the vendors are pressured to push their AI tools, but anything people show off are barely functional low/no code solutions. Usually it handles a small subset of their prior automation solutions, with vague promises that they’ll extend functionality and won’t cost an order of magnitude more.

[–] yogthos@lemmy.ml 8 points 17 hours ago* (last edited 5 hours ago) (1 children)

I've been using opencode for actual projects at work. DeepSeek v4 can code up a lot of stuff fairly confidently. If you give it clear requirements, tell it to make a phased plan, use TDD, and commit after each phase, it tends to produce decent code. I just do code reviews against the diff and then tell it to fix anything I don't like. It's also great for spelunking through large codebaes. You can easily trace through how an endpoint works for example, get it to write stuff like sample curl queries, etc.

But the thing is that you don't actually work all that much faster. You still have to review everything. You have to actually manually use the app and make sure it works functionally. Like you basically can't trust anything it does. So, it makes my life easier. I don't have to look up API docs, figure out how random libraries work, or having to write a bunch of boilerplate. But it doesn't replace me, and it doesn't actually result in me working significantly faster.

[–] queermunist@lemmy.ml 5 points 15 hours ago (2 children)

The dream is deskilling, not necessarily making workers faster or more efficient. Instead of relying on highly skilled employees who have tons of leverage because of their experience and education, they'd be able to just put anyone on the keyboard to review and troubleshoot the chatbot's outputs. They essentially want coding to require as much skill as any other assembly line work.

[–] yogthos@lemmy.ml 5 points 8 hours ago* (last edited 5 hours ago)

I mean they want that, but it's just not happening with the way the tech works right now. Unless you have a deep understanding of the problem you have the model solve, then you have no way to evaluate whether it solved it correctly or not. And it's basically like an evil genie where it will interpret your request in a dumbest way possible by default. So, you get decent results when you already know what the shape of the solution should be, and you give the model concrete direction on the approach to take, algorithms to use, and so on. And that's why the whole idea of deskilling workers or replacing them with automation is not really working out. You'd need genuine artificial intelligence for that and LLMs aren't it.

[–] gary_host_laptop@lemmy.ml 2 points 8 hours ago

that's still not possible though, you need to know what to ask, you need to have some idea of how programmes work, etc.