AI coding tools are optimising the wrong thing and nobody wants to hear it. Writing code was already fast. The bottleneck is everything else: unclear requirements, review queues, terrified deploy cultures, and an org chart that needs six meetings to decide what colour the button should be.
Not to mention that having AI fulfill the responsibilities of junior devs means than in the long run, you’re preventing junior devs from growing into senior devs.
Eventually we’re going to have a problem where no one is qualified to know what the hell is going on.
In my experience, the biggest problem with AI-written code is that is is full of hidden bugs that are near impossible to track down, because the code is virtually unreadable. Basically, you deploy faster and spend a lot more time with catastrophic failures you have to root out, only to find out that the fixes to the issues cause more issues downstream, because nobody know what anything does anymore.
… am i alone in reading this and getting that queasy twinge in the back of my mind that “hey this writing feels like it has some LLM-like patterns…”
Something in the formatting. Something in the phrasing. Something in the “It’s not x. It’s y.”
At least it’s not full of emdashes, and the phrase “let’s unpack this” makes no appearances.
i wish the tsunami of slop hadn’t made me so damn paranoid -_-
You know the face - it’s the one where they’re calculating whether to say something or just update their LinkedIn later.
yeah, it’s ai slop complaining about ai slop
I also got LLM vibes. However, some humans do just genuinely write like that. It’s particularly an issue with ESL speakers getting caught in false positives, although this author seems to be a white Australian guy who is probably a native English speaker.
I suppose if you were really bothered you could go back and look at his writing before the dawn of the vibes and see if his writing style is about the same. I don’t care enough to check.
Every move of the past 20 years has been to deploy code faster, and all it cost us was several hundred times the memory and CPU usage to the point where you need 32 Gb of RAM to open two web pages.
It’s just as stupid as it was when it started.
Write imperative code.
Imperative code has nothing to do with any of that.
Modern frameworks like React and its ilk make up 99% of the overhead and inefficiency of modern web code, and that is all declarative boilerplate. Things that should be one imperative line of code operating on a single html element turn into complete recomputing of every element in a thousand element DOM triggered not by direct invocation, but a global state change.
I won’t get into too much of a rant about this, but I’m in DevOps but I used to be a full-stack dev (as in doing nothing well).
My UI lead is just SO proud that he doesn’t need to write any code by hand and can show you how to do it too! Guess what he doesn’t do? Doesn’t fix up or speed up the website. We integrate with two of our sister products. One of them by sharing our UI packages through our private NPM repo and the other by iframe. Both are done so poorly at a fundamental level because our NPM packages are just SO poorly put together that if you spent any time looking at the output, you’ll see that it’s like typescript but not really. It’s not CJS nor does it really do ESM. Our sister websites needs to use another bundler to import these packages.
Our website is just slow too. Forget about deferred stuff or tighter bundles or reducing the page load times … just putting out more with AI is the key now. Let’s just forget about fundamentals.
the consequences of 20 years of “move fast and break things”: people have to replace perfectly good hardware after 5 years to keep up, and everything’s so broken the literal rocket scientists can’t figure it out.
It is said in the development community that as a junior you write more code and think less, and when you become more senior you think more and write less code.
Experience leads to the technical equivalent of ‘measure twice, cut once’.
Exactly, and now we’re rushing through the important stuff because we’re forced to output more due to the “productivity” of AI.
Don’t forget thinking…
