washedup is right, of course, that the community is many people with many views (though the point should be made less snarkily). Of course HN disagrees with itself; it's not a self to begin with.

I think there's another factor: it takes time to find the best ways of using such new tools. The community is working this out slowly (as is society generally) and it is naturally a lagging phenomenon.

Seeing generated sites like OP and karpathy's HackerTube (https://news.ycombinator.com/item?id=50319510) which look good and don't have the obvious-LLM sameiness that so many others do, is an interesting step along this way. It's also refreshing. There's no intrinsic reason why generated software should look bad or work poorly.

It's not about the tools.

I also regularly complain about AI generated content. I do that mostly, because it's often either

- extremely uninspired design and UX wise (e.g. the classical beige/serif AI slop website),

- full of obvious and stupid mistakes (e.g. those shitty infographics on social media),

- just mimics the real thing, but misses the actual purpose (e.g. again those shitty infographics on social media)

- or is ridiculously overloaded with features and documentation (e.g. many AI generated software tools on GitHub)

That's repelling to me, because it screams that the human behind it is himself lazy and uninspired. It's bland, overloaded or full of mistakes, because raw output is a commodity.

It feels like a waste of time to even consume something, that the author or developer wasn't willing to put actual effort and thought into.

But that's an artefact of how the author is using the tool and presenting the results.

It's not an inherent property of AI use itself. HackerTimes for example doesn't feel uninspired or bland. And i like it.

It's almost as if there are different groups that have different views and upvote different things.

Goomba.