It is not possible to filter ai content from search results, even duckduckgo links to articles shat out by some mindless fucking toaster. I’ve had enough of it.
It’s only a matter of time until the ever flowing diarrhea deluge of “AI” generated information will engulf the internet. Soon all information online will be useless or might as well be because it will be impossible to tell the difference.
I think it’s wise to download a full dump of Wikipedia and other sources. Preferably from before AI.
Are there any other websites or information packages worth getting? Are there hosts that aren’t contaminated with this digital disease?
You must log in or # to comment.
deleted by creator
Some time ago I downloaded dump of all stackexchange sites. It’s not deployed to my home lab, just sitting there on hard drive so that I know I have it even when Internet goes DataKrash.
I cannot recall which dump I have. I believe it’s sqllite DB from 2024. I think newer dump are xml only. Here is good starting point - https://meta.stackexchange.com/questions/224873/all-stack-exchange-data-dump-releases
Look into Kiwix if you haven’t. Tons of educational and informational content that you can download for offline use. Everything opens in a browser of sorts and is searchable.
Also, if there is a topic you are interested in, buy books, download books. My wife and I started building quite the library when certain political groups started trying to ban books. The acceleration of AI has kept us going.
Consider this
https://marginalia-search.com/
I haven’t looked into it much yet, but it looks liker a step in the right direction
The internet was doomed once amazon links became a thing.
If I’m searching for something that existed before 2021 then I limit the date range of my searches to before that time. DDG lets you do this in the search options.
Sometimes you click a result from before 2021 but the website has been updated more recently (a lot of articles have been re-written with LLMs since 2024) in which case I load the page in the waybackmachine to see what it looked like before.
This combination has worked quite well for finding authentic results. If you’re searching for something that requires recency then you’re SoL.
