otherayden

I found it bc I started getting new traffic to the site from HN and turns out someone else was actually plugging it in another comment here

I run https://www.unslop.news/ which uses an LLM (ironically) to filter out content discussing AI. There's text at the bottom to indicate what fraction of the first 3 pages gets filtered out, and it's pretty much always at 50% or more.

Although there is some irrational hysteria about the whole "chatgpt is drinking all our water" thing, to me the biggest issue (other than the mass piracy) is local energy bills going up without contributing to the local economy. Seems like it would do more harm than good for all but the owners of the datacenters.

I appreciate the way you sum up the ML vs traditional software appeal and I think it's helped me articulate why ML is so unappealing to me. The power of software has very little to do with the power of the machine it's run on (with some obvious exceptions like games and physics sims) which made it very exciting to me when I was learning to code. The same cannot be said about deep learning where more data and more compute means more power.

Consider trying https://unslop.news/ , I tried to solve this exact problem by classifying post title and webpage contents as AI or not and filtering out the AI stuff. Also, tangentially, if you're looking for a really impressive technical writeup I recently enjoyed this one :) https://philo.gay/linecam/

Yeah, I like the way lobste.rs does it with the vibecoding tag

I made a similar site https://unslop.news/ but rather than keywords it actually feeds the title and article to an AI which classifies it as "about AI" or not to filter out the AI discussion. Also it's open source: https://github.com/leiDnedyA/hn-without-ai . Hope it helps!

This headline would mean something very different 10 years ago lol

We need to bring back the days when this stuff hit the front page more often!

very impressed by the images you got, zero choppiness

These are so adorable!

Same here. I'm especially a big fan of the "using x as y" format (using youtube as a mountable file system, using my wii to host a web server, etc). I feel like they're a lot less common on here these days than a year or 2 ago

edit: I made this site to only show non-ai related posts like this one https://www.unslop.news/

Hi I updated the title because the previous one was a little misleading. I'm using GPT 5.6 sol and simply asking if the content and title indicate that the article is primarily focused on AI, rather than being generated by AI. I checked hugging face and found surprisingly few models dedicated to detecting AI-generated text though, but there are some APIs and sites like GPTZero available for it. Perhaps the "detect AI-generated content and remove it from feeds" problem could be an interesting one to investigate though

Yeah sorry for that, I updated the title. Seems like filtering out AI-generated content from feeds would be an interesting problem to attack though

Thank you for the rec, I used it to identify a mystery house plant I was gifted :)

By "read content in their second language", do you mean "read content originally authored in their second language" or "read content auto-translated to their second language for the sake of language practice"?

I took a crack at implementing this, it goes a little further by classifying with an LLM (both the title and the article content if it can access it)

https://hn-without-ai.vercel.app/

I tried this experiment out and it is right on the money. The ad I got was AI generated video with AI script & audio selling fake 16TB SSDs. Wonder if anybody has some insider info on how these ads are produced at scale / who's behind them

Huge meld fan here, recommended by a professor about a year ago. Game changer

And that Optifye.ai demo with the sweatshop surveillance software

I think we kiss of deathed the article haha. Here's an archive https://archive.is/6QD8c

It seems to vary a lot by the day. HN still has enough good days so I keep checking, but on the bad days I just go to lobste.rs and usually there’s something quality to read over there

To speed it up, run

['whiteBall', 'blackBall'].forEach(color => { data[color].v.x *= 5; data[color].v.y *= 5 });

In dev console :)

Let’s focus specifically on aliens that exist somewhere in the universe, regardless of intelligence. My one criteria is that if it was seen by an average human that they would think “yep, this is definitely life”

I’ve actually been working on a tool that uses jina.ai’s “reader” api to let you see a rendered markdown view of a pages contents. Check it out at https://leidnedya.github.io/markweb or try it out with https://leidnedya.github.io/markweb#example.com

Would you be willing to share usage data on this? This is an interesting case of LLM products seeming really useful, but I’m wondering if there’s a big market for this

I wish I had a mac to build this for you, sounds like a fun project

This seems like such a no-brainer, I’m shocked this isn’t already something sites need explicit permission to do

thank you! I love HN and really hope people here get some value from my thing!

yeah, it's a pretty incremental change but it will find the latest archive of the link, and as long as it exists, you will be redirected to it. If there's no archive yet, it automatically puts you in the archive.today queue to get your page archived