←back to thread

468 points bookofjoe | 1 comments | | HN request time: 0s | source

I would very much like to enjoy HN the way I did years ago, as a place where I'd discover things that I never otherwise would have come across.

The increasing AI/LLM domination of the site has made it much less appealing to me.

Show context
jxjnskkzxxhx ◴[] No.44574698[source]
Weird that people are floating the idea of kicking out of a tech forum the most important tech development of the last 10 years.

Not sure what that means about the community, but must mean something.

replies(4): >>44574853 #>>44574867 #>>44575360 #>>44576880 #
zahlman ◴[] No.44574853[source]
The problem is the quality, not the topic. Understanding serious papers about AI development requires fairly specialist knowledge; there are plenty of people around (like myself) who have been programming for decades and can write really nice code in a bunch of different programming languages, but have very little if any mental model of "transformers" or whatever.

So in practice, "AI" content ends up revolving around people bandying about opinions about whether or not we're all doomed, or whether or not we're all on the edge of a utopia, or how much productivity programmers (and which ones) have lost or gained, or what kinds of tasks the LLMs are or are not currently or still good at, or whether anyone still cares about the fact that the term "AI" is supposed to mean something broader than LLMs + tool use.

The emergence of the "vibe coding" concept has made things worse because people will just share their blog posts about personal experiences with trying to write code that way, or flood the Show HN section with things that are basically just "I personally found this specific thing to be 'the boring stuff' that's actually relevant to me, so now I'm automating it" with a few dozen lines of AI-generated code that perhaps invokes some API to ask another AI to do something useful.

replies(1): >>44574916 #
jxjnskkzxxhx ◴[] No.44574916[source]
Interesting take.

To me it feels like golden age of hackers in the 60s-80s (which was before my time but I heard stories about) where everybody is doing their own home grown research to the best of their abilities and sharing insights of varying quality.

But somehow these days if it's not all polished, HN "hackers" aren't interested.

replies(2): >>44575157 #>>44575174 #
1. zahlman ◴[] No.44575157[source]
> But somehow these days if it's not all polished, HN "hackers" aren't interested.

The fun part is that these days, typically the READMEs (especially) and licensing and documentation and maybe even the packaging setup are "polished"; the actual code (and perhaps the tests), not so much. It's quite backwards from what you expect from humans writing new code based on personal intrinsic motivation.