←back to thread

Nobody knows how to build with AI yet

(worksonmymachine.substack.com)
526 points Stwerner | 4 comments | | HN request time: 0.001s | source
Show context
aprilthird2021 ◴[] No.44616732[source]
Eh, idk. First of all, the article is really wordy to say very few things. That just frustrated me a bit.

Second of all, it's easy to fart out some program in a few days vibe coding. How will that fare as more and more features need to be added on? We all used to say "Dropbox that's just FTP wrapped in a nice UI anyone can make that". This protocollie project seems to be a documentation viewer / postman for MCP. Which is cool, but is it something that would have taken a competent dev months to build? Probably not. And eventually the actual value of such things is the extensibility and integrations with various things like corporate SAML etc.

Will the vibe code projects of today be extensible like that, enough to grab market share vs the several similar versions and open source versions anyone can make in a few days, as the author suggests? It can be hard to extend a codebase you don't understand because you didn't write...

replies(3): >>44616771 #>>44617126 #>>44617171 #
1. ModernMech ◴[] No.44617126[source]

  You prompt. You go live your life. You come back to ten thousand lines of code. You spend 5 minutes reading. One sentence of feedback. Another ten thousand lines appear while you're making lunch.
Yeah, it strikes me the author writes prose the same way they're generating code. 20k lines? That's enough code for a whole compiler or an operating system kernel. I'd love to see what those 20k lines actually do -- notably, in these articles about AI, people tend to not link the actual code when they easily could, which is curious. I mean, my macro expander can also write 20k lines of code while I eat lunch, but no one is pretending it's sentient and about to replace devs.
replies(1): >>44617255 #
2. wilkystyle ◴[] No.44617255[source]
I definitely did a double take when I got to this section. I am neither an AI optimist nor an AI pessimist (probably slightly on the optimistic side of the midpoint) but this sounds insane to me for any software that people might truly depend on. Five minutes of review for 10,000 lines, happening multiple times per day?
replies(1): >>44617469 #
3. ModernMech ◴[] No.44617469[source]
I did some digging, and this seems to be the project the author is using AI to work on: https://github.com/sublayerapp/sublayer

You look at the PRs... there are 786(!) AI generated pull requests and an associated AI generated code review for each one. Each PR is about ~20-100 lines of Ruby (including comments) that implements an "action" for the sublayer system as a Ruby class. So probably something that could be handled by a macro expander. Or at least it's AI used as a fancy macro expander.

But yeah, there's about 20k lines of code right there easily. Although, because it's Ruby, it's not (much) of an exaggeration to say ~50% of the generated lines are a single "end" keyword.

The author is someone who before AI, would publish ~300 commits a year to Github. This year they are on track for 3000 commits using AI. But the result seems to be that PRs are accumulating in their repo, implementing hundreds of features. I'm wondering why the PRs are accumulating and not getting merged if the code is good? Is the bottleneck now review? What would happen if AI took over PR merging as well as PR creation?

replies(1): >>44644111 #
4. thrawa8387336 ◴[] No.44644111{3}[source]
He deleted the PRs? Struck a nerve