Hacker News Reader

Comment by hparadiz | Hacker News Reader

I'm like two months into a vibe coded C project. My issues are the same as ever. How to pack memory. What syscalls to run and when. Is the program stable after running for 24 hours? When I want to make a change it's usually a trade off with something else. There's no accounting for taste among humans. Let alone among an LM. It's great at implementing my ideas but terrible at coming up with those ideas. Architecture is always going to be king.

View on HN

I agree. Humans with experience are better at writing and designing code. But for the people who have given up on themselves this stuff is interesting.

I personally use these models for low value boiler plate tasks only. Or auto complete.

I meant literally vibe coded. It's 99.99% written by agents. But it's extremely opinionated on the architecture. How the event bus works. How the plugin ABI is structured. What syscalls to run and when. That's my opinion. The human being. Outside of that the whole thing may as well be a compiler writing assembly. I'm actually thinking of doing that as my next experiment instead of relying on gcc.

Models are heavily fine tuned and trained to follow instructions. They are trained to be subservient. I am sure that cuts into their ability to think creatively. The other risk with a lot of creative thinking is risking hallucinations (creative thinking = perhaps trying what’s not in its training set = hallucination basically). So I will rephrase creative thinking as desired or useful hallucination that is still firmly within the constraints of the prompt.

If that sounds complicated, that’s because it is! It’s a tricky balance to get right. I think the current architecture for most GPT models isn’t sufficient to solve this problem for good. I suppose we need to do more research into what constitutes desirable vs undesirable hallucination and how to shift the balance towards the latter.

Reply on news.ycombinator.com