Hacker News Reader

Comment by vova_hn2 | Hacker News Reader

> yeah it kinda works but not really well enough

I mean, at some point it was true.

I remember that around 2023, when I first encountered colleagues trying to use ChatGPT for coding, I thought "by the time you are done with your back-and-forth to correct all the errors, I would have already written this code manually".

That was true then, but not anymore.

View on HN

This is true for things you already understand. It works for implementing yet another CRUD view because I've done it a million times before. I know exactly what the code should look like, but it takes a while to type it in. When my typing speed is the bottleneck then of course LLMs win (and I use them for that all the time).

But the interesting stuff where you don't understand the problem yet, it doesn't make it quicker. Because then the bottleneck is my understanding. Things take time. And sleep. They require hands-on experience. It doesn't matter how fast LLMs can churn out code. There's a limit to how fast I can understand things. Unless, of course, I'm happy shipping code I don't understand, which I'm not.

Unfortunately, it seems that few of the people willing to pay money for software development actually care whether you understand the code or not. Unless it breaks in a publicly embarrassing way, of course. Then "you own the code" and the sacrificial human in the loop can be dredged out to take the blame.

All the incentives nudge you towards less and less critical evaluation of the output. The results of careful evaluation are much harder to measure than "this guy is cranking out 10x the code compared to last year!" And while you were busy thoughtfully internalizing the output, the guy next to you has been letting Strega Nona's pasta pot go brrr and spew another thousand lines of spaghetti on top, ready for you to review. Eventually "lgtm" becomes the default, and "do you want me to go ahead and implement that for you" starts sounding like the only way to keep your head above water.

I think it's still true, but very domain specific. I am not confident it will stay true.

No, it's still very much true. Every now and then I use an LLM to write code and the vast majority of the time it turns out to take just as much time (if not more) than it would've taken to write the code myself.

Exactly. Verification is not cheap at all

You are either using it wrong or you are writing extremely niche code that has bad llm coverage

or you are in denial about what he is saying

I suspect I fall into the former camp, but I'm not sure where to start when it comes to learning how to use llms "the right way".

I'm not a proper software engineer, but I do a lot of scripting and most of my attempts to let a model speed up a menial task (e.g. a small bash or python script for some data parsing or chaining together other tools), end up with me doing extensive rewrites because the model is completely inconsistent in naming convention, pattern reusage, etc.

Reply on news.ycombinator.com