Using paid search engine #Kagi for months now. At first it seemed a bit much to pay $10 a month for what I thought was a more private #Google wrapper.
Now I can safely say its worth the money.
The results are way cleaner, and with the power to fine tune results to my needs (e.g. globally block/lower junk sites).
Say what you want about "AI" but the optional LLM-based quick-answer is one of the most simple, useful, ethical, and accurate LLM integrations I have ever seen.
[#]AI #search #LLM
=> More informations about this toot | More toots from vjprema@fosstodon.org
[#]Kagi philosophy on AI integration also gives you some perspective.
[#]LLM features should ideally always be optional and serve to enhance not obfuscate nor create dependencies, nor pretend to be a "real" answer with no user control/transparency.
Sources and provenance should be shown where possible.
All of this is possible when an LLM is used within its current limits: to transform/summarize correct answers from a RAG (search results), rather than generate answers.
https://help.kagi.com/kagi/why-kagi/ai-philosophy.html
=> More informations about this toot | More toots from vjprema@fosstodon.org
@vjprema very cool. I've been using Brave search for a few months and I like it, but if I need something else I'll keep Kagi in mind.
=> More informations about this toot | More toots from nathanj@fosstodon.org This content has been proxied by September (3851b).Proxy Information
text/gemini