It is becoming near impossible to find relevant information from search engines. Duckduckgo, SearXNG, Bing, Google, and so many more mainstream engines have a significantly high noise to signal ratio, and it is getting worse.
Here are a collection of the best search engines I know, please add more to the list.
- Forum Search Engine: https://crowdview.ai/
- Non-commercial Search: https://search.marginalia.nu/
- Libre Meta Search Engine: https://librey.devol.it/
- Golden Age Search Engine: https://www.wiby.org/
- Yandex: https://yandex.com/
If no more high quality search engines exist, would it be possible to host your own?
EDIT: Some new discoveries. The addon uBlacklist and filters can block super SEO sites from appearing in search.
I’ve found bing ai is quite good if you ask for the source after anything it spits out.
These models can invent a source. Their only incentive is to have a convincing conversation with you. They are unconcerned with the truth.
What I mean is I use it to get the links to those sources. Like when you use Wikipedia as a jumping off point. I don’t think we’re at the point yet where we have the problem Wikipedia sometimes has that the sources used sometimes themselves just cite Wikipedia.
The links to Wikipedia are actual citations to real sources. LLMs basically just generate something that looks like the link to a credible source which might support what it’s said. It doesn’t care if its “source” actually supports what it says.
I read an interesting article a few years ago about the Wikipedia source problem. It did a dive into how sources that seem legitimate on Wikipedia can and up citing sources that are less so. They were able to trace back the citations to Wikipedia itself. So no, they’re not always real sources.
Which is why you read the page it has linked for you as a source. Unless you’re trying to say it full on generates a page for you.