←back to thread

Nobody knows how to build with AI yet

(worksonmymachine.substack.com)
526 points Stwerner | 4 comments | 19 Jul 25 15:45 UTC | HN request time: 0.648s | source
Show context
fmbb ◴[19 Jul 25 16:11 UTC] No.44616724[source]▶
That time dilation feels a bit like what METR reported:

https://metr.org/blog/2025-07-10-early-2025-ai-experienced-o...

Developers believe they complete tasks 25% faster with AI but when measured they are 19% slower when using AI.

replies(4): >>44616759 #>>44616807 #>>44617418 #>>44622100 #
logicchains ◴[19 Jul 25 16:20 UTC] No.44616807[source]▶
It depends on how the AI is used; there's a huge difference in productivity between a structured workflow with repeated automated feedback to the AI, and just ad-hoc prompting. For instance, Gemini 2.5 coding up a spec-compliant HTTP2.0 server in two weeks: https://outervationai.substack.com/p/building-a-100-llm-writ... . 15k lines of code and 30k lines of tests; no human coder could produce something like that so fast.
replies(3): >>44616883 #>>44617034 #>>44622197 #
djeastm ◴[19 Jul 25 16:29 UTC] No.44616883[source]▶
Would that application not exist in the training data already?
replies(1): >>44617046 #
1. logicchains ◴[19 Jul 25 16:46 UTC] No.44617046[source]▶
The are around three Golang HTTP2.0 servers on Github, but it wasn't just regurgitating the application from memory, as if it was then it would have been mostly correct first try, and it wouldn't have needed to spend 80%+ of the development time in a code-compile-test cycle of fixing bugs identified by integration tests and spec conformance tests.
replies(1): >>44617110 #
2. codingdave ◴[19 Jul 25 16:51 UTC] No.44617110[source]▶
So you are saying that LLMs have no capacity to re-use existing work, and will just burn hours and money re-inventing wheels?
replies(2): >>44617367 #>>44620660 #
3. bavell ◴[19 Jul 25 17:18 UTC] No.44617367[source]▶
Welcome to The Future™!
4. simonw ◴[20 Jul 25 00:05 UTC] No.44620660[source]▶
If you tell it to reinvent the wheel, it will.

If you tell it to use existing libraries (and they are in its training data) it will do that instead.

If you tell it about libraries it hasn't encountered before it can use those instead.