World News

Should cars have been outlawed because it put farriers and stables out of business? Were shipping containers a bad idea because they required fewer longshoremen?

Technology comes and makes jobs obsolete. It happens all the time. It just happens that this technology has come in a big, visible way, and many of the ways its used and marketed are useless and/or awful. That doesn't mean it's entirely bad, and there's certainly no way to stop it now.

AI will replace jobs. We can't get around that fact. Companies that fail to adapt will fall behind. Whether I use it at my job or not has no bearing on the industry, and I'm not in a position to push for industry-wide change (nor is the company I work for). So we can either use it, or also fall behind.

I work for a mid-sized company. We still hire junior developers. I don't think we have any plans to get rid of them entirely, but I'm not involved in that process. But after a couple decades of huge growth in the industry, developers (especially junior ones) are going to have a rough few years as the industry realigns with the new normal. There will be job losses, there will be companies that disappear entirely because they either depended too much upon AI, or didn't adapt fast enough. But pretending that AI isn't a useful tool when used in specific ways is just sticking your head in the sand.

In my experience one needs to be a senior developer with at least some experience with their own code having gone through a full project lifecycle (most importantly, including Support, Maintenance and even Expansion stages) to really, not just intellectually know but even feel in your bones, the massive importance in reducing lifetime maintenance costs of the very kind of practices in making code that LLMs (even with code reviews and fixes) don't clone (even when cloning only "good" code they can't do things like for example consistency, especially at the design level).

• Inexperienced devs just count the time cost of LLM generation and think AI really speeds up coding.
• Somewhat experienced devs count that plus code review costs and think it can sometimes make coding a bit faster
• Very experience devs looks at the inconsistent multiple-style disconnected mess (even after code review) when all those generated snippets get integrated and add the costs of maintaining and expanding that codebase to the rest, concluding that "even in the best case in six months this shit will already have cost me more time in overall even if I refactor it, than it would cost for me doing it properly myself in the first place".

It's very much the same problem with having junior developers do part of the coding, only worse because at least junior devs are consistent and hence predictable in how they fuck up so you know what to look for and once you find it you know to look for more of the same, and you can actually teach junior developers so they get better over time and especially focus on teaching them not to make the worst mistakes they make, whilst LLMs are unteachable and will never get better plus they're mistakes are pretty much randomly distributed in the error space.

You give coding tasks in a controlled way to junior devs whilst handling the impact of their mistakes because you're investing in them, whilst doing the same to an LLM has an higher chance of returning high impact mistakes and yields you no such "investment" returns at all.

can work for a few lines, for some boilerplate, or for some refactoring

I highly doubt the person you're replying to meant anything else. We're all kinda on the same page here.

Oh, you wanted code that actually works with your existing codebase? Sorry, here's some additional sludge sprinkled on top of the previous crappy answer.