Hacker News Reader

Comment by p2detar | Hacker News Reader

> Yes, AI can generate code quickly (whether that’s a good thing is open for debate), but that doesn’t mean it’s generating the correct code.

No, the code is actually almost always correct. The way it’s added is probably not what you’re going to like, if you know your code base well enough. You know there’s some ceremony about where things are added, how they are named, how much comments you’d like to add and where exactly. Stuff like that seems to irritate people like me when not being done right by the agent, and it seems to fail even if it’s in the AGENTS.md.

> If you were to give human developers the same amount of feature/scope documentation you would also see your productivity skyrocket.

Almost 2 decades in IT and I absolutely do not believe this can ever happen. And if it does, it’s so rare, it’s not even worth talking about it.

View on HN

>No, the code is actually almost always correct

That's not my experience, especially when the inputs are bugs or performance issues. It frequently hallucinates and misdiagnosis without a guiding hand. However, it can still RCA and analyze well and improve efficiency if you keep an eye on what it's doing and push it the right direction.

> If you were to give human developers the same amount of feature/scope documentation you would also see your productivity skyrocket.

I think you run into a ceiling how fast a person can digest and analyze the info compared to a machine

What tools are you using? What settings? What process? What's your code review like?

I think this varies a lot. I find with a c++ project I'm working on that the LLM needs a lot of guardrails and guidance, and still gets a lot wrong. But with a vite/js project it often one shots complex and intricate changes in large codebases.

Claude Code with Opus 4.7 xhigh effort

Datadog + Python with a little AWS

It does better at writing code but diagnosing bugs and performance issues is a bit of a crapshoot. It's decent at diagnosing but when tasked with improving performance it gets a bit sketchy. It'll write something that runs but often times it either has subtle functionality changes or mediocre improvements

Other times I'll repeatedly have to push it to find the root cause. It'll say "hey this downstream service threw an exception that propagated as a 500" and I'll have to remind it that's a first party service with info available in the trace and source code also available that it needs to continue investigating.

> No, the code is actually almost always correct.

I wouldn't go that far. It is actually almost always incorrect in that its incomplete. It also needs more iterating and bug fixing.

> Almost 2 decades in IT and I absolutely do not believe this can ever happen. And if it does, it’s so rare, it’s not even worth talking about it.

It happens completely routinely, IME. Just compare how much effort it takes to clone a system that's already written versus making that system from scratch.

Reply on news.ycombinator.com