Hacker News Reader

Comment by hmaxwell | Hacker News Reader

I feel like people who can't get AI to write production ready code are really bad at describing what they want done. The problem is that people want an LLM to one shot GTA6. When the average software developer prompts an LLM they expect 1) absolutely safe code 2) optimized/performant code 3) production ready code without even putting the requirements on credential/session handling.

You need to prompt it like it's an idiot, you need to be the architect and the person to lead the LLM into writing performant and safe code. You can't expect it to turn key one shot everything. LLMs are not at the point yet.

View on HN

That's just the thing though - it seems like, to get really good code out of an LLM, a lot of the time, you have to describe everything you want done and the full context in such excruciating detail and go through so many rounds of review and correction that it would be faster and easier to just write the code yourself.

Yes, but please remember you specify the common parts only once for the agent. From there, it’ll base its actions on all the instructions you kept on their configuration.

Welcome to the waterfall development model. This is what companies did before enshitiffixation

I’ve found LLMs to be severely underwhelming. A week or two ago I tried having both Gemini3 and GPT Codex refactor a simple Ruby class hierarchy and neither could even identify the classes that inherited from the class I wanted removed. Severely underwhelming. Describing what was wanted here boils down to minima language and they both failed.

I tried getting AI to update some JUnit 4 to Junit 5 - it replaced the JUnit 4 assertions with Java's built-in assert keyword. Very underwhelming.

Exactly this. Not sure what code other people who post here are writing but it cannot always and only be bleeding edge, fringe and incredible code. They don't seem to be able to get modern LLMs to produce decent/good code in Go or Rust, while I can prototype a new ESP32 which I've never seen fully in Rust and it can manage to solve even some edge cases which I can't find answers on dedicated forums.

I have a sneaking suspicion that AI use isn't as easy as it's made out to be. There certainly seem to be a lot of people who fail to use it effectively, while others have great success. That indicates either a luck or a skill factor. The latter seems more likely.

What are your secrets? Teach me the dark arts!

There are wide gaps in:

1) the models people are using (default model in copilot vs. Opus 4.5 or Codex xhigh)

2) the tools people are using (ChatGPT vs. copilot vs. codex vs. Claude code)

3) when people tried these tools (e.g., December saw a substantial capability increase but some people only tried AI this one time last March)

4) how much effort people put into writing prompts (e.g., one vague sentence vs. a couple paragraphs of specific constraints and instructions)

Especially with all the hype, it makes sense to me why people have such different estimates for how useful AI actually is.

This sounds like my first job with a big consulting firm many years ago (COBOL as it happens) where programming tasks that were close to pseudocode were handed to the programmers by the analysts. The programmer (in theory) would have very few questions about what he was supposed to write, and was essentially just translating from the firm's internal spec language into COBOL.

I find that at the granularity you need to work with current LLMs to get a good enough output, while verifying its correctness is more effort than writing code directly. The usefulness of LLMs to me is to point me in a direction that I can then manually verify and implement.

Reply on news.ycombinator.com