Ask HN: Do LLMs really code better than you?
I keep seeing this sentiment in the comments on here from supposed 25 year storied career programmers.
I’ve been programming for 14ish years professionally and can say for a fact an agent doesn’t code better than I do. I’ve used the latest models and even with heavy steering the code it produces is not subjectively or un subjectively “better”. It is more verbose, more defensive, has more comments, it’s done faster.
The fact someone I highly respect that uses the agentic workflow effectively told me to “stop being so precious about the code” makes it obvious to me that both can’t be true.
So please, explain to me, what are you guys talking about?
(
no comments
| on HN
)
There are no comments yet.