Live data from Hacker News

After months of coding with LLMs, I'm going back to using my brain

albertofortin.com

11–20 of 229 posts

Re: After months of coding with LLMs, I'm going back to using my brain

#11
I get it and I see the same problems as the author.

I'm working on a few toy projects and I am using LLM for 90% of it.

The result is 10x faster than if I coded it "by hand", but the architecture is worse and somewhat alien.

I'm still keeping at it, because I'm convinced that LLM driven code is where things are headed, inevitably. These tools are just crazy powerful, but we will have to learn how to use them in a way that does not create a huge mess.

Currently I'm repeatedly prompting it to improve the architecture this way or that way, with mixed results. Maybe better prompt engineering is the answer? Writing down the architecture and guidelines more explicitly?

Imagine how the whole experience will be if the latency was 1/10th of what it is right now and the tools are 10x better.

Re: After months of coding with LLMs, I'm going back to using my brain

#12
I prefer to ask LLMs to help me to create highly cohesive functions, and absolutely abhor using agents with their overly verbose code, their tendency to do shotgun surgery and the inevitable regressions.

So I kind of do a top-down design and use the LLM to help me with toil or with unfamiliar things that would require me finding the right documentation, code examples, bug reports, etc, etc... LLMs with web search are great for this kind of toil elimination.

LLMs are useful tools, but they have no notion of hierarchy, of causal relationships, or any other relationship, actually. Any time they seem to have those capabilities in the code they generate, it is merely a coincidence, a very probably coincidence, but still, it is not intentional.

Re: After months of coding with LLMs, I'm going back to using my brain

#13
post #8

I don't get the whole "all-in" mentality around LLMs. I'm an iOS dev by trade, I continue to do that as I always have. The difference now is I'll use an LLM to quickly generate a one-off view based on a design. This isn't a core view of an app, the core functionality, or really anything of importance. It's a view that promotes a new feature, or how to install widgets, or random things. This would normally take me 30-…

People just do stupid stuff like "going all in" for their blog posts and videos. Nuance, like rationalism, doesn't get engagement.

Re: After months of coding with LLMs, I'm going back to using my brain

#17
> One morning, I decide to actually inspect closely what’s all this code that Cursor has been writing.

You can't abdicate your responsibility as a builder to the LLM. You are still responsible for the architecture, for the integrity, for the quality. In the same way you wouldn't abdicate your responsibility if you hired a more junior engineer.

Re: After months of coding with LLMs, I'm going back to using my brain

#19
> So I do a “coding review” session. And the horror ensues.

Yup. I've spoken about this on here before. I was a Cursor user for a few months. Whatever efficiency gains I "achieved" were instantly erased in review, as we uncovered all the subtle and not-so-subtle bugs it produced.

Went back to vanilla VSCode and still use copilot but only when I prompt it to do something specific (scaffold a test, write a migration with these columns, etc).

Cursor's tab complete feels like magic at first, but the shine wore off for me.

Re: After months of coding with LLMs, I'm going back to using my brain

#20
Coding and writing is like arithmetic and algebra, use it or lose it. You don’t gain fluency by throwing everything into Mathematica.

Of course the real question is, is there any reason to be good at coding and writing, if LLMs can just do it instead? Of course it’s hard to sell that we should know arithmetic when calculators are ubiquitous.

Personally, I value being skilled, even if everyone is telling me my skills will be obsolete. I simply think it is inherently good to be a skilled human

Post reply on HN