Hacker News Reader

Comment by TriangleEdge | Hacker News Reader

Does Kagi have any value in the era of LLMs? My understanding is that it aggregates result from different providers.

Yes: you get reliable source information and don’t get inaccurate summaries. E.g. last week I used Gemini to answer a plant biology question and got two contradictory answers based on minor variations in the wording because it incorrectly relied on blog spam over peer-reviewed articles for the first query.

The initial false answer was baldly asserted by the LLM without sources in the first two paragraphs but some of the phrasing it used was enough to locate the non-authoritative blog content it was apparently laundering. Had it accurately cited sources, it would’ve been easy to see that this random WordPress site saying X wasn’t as authoritative as the PubMed hits saying !X.

You say that as if LLMs are a good thing.

You say that as if a technology can be easily classified as good or bad.

Touché.

Still, I think the down sides of LLMs outweigh the benefits for most use cases.

Kagi assistant is effectively a superset of other LLM chat apps.

Has access to kagi search which is a also a superset of search backends for the assistant

I second the utility of the Kagi Assistant. I didn't think I would use it much but now do so constantly. Especially because ending a regular search query in a question mark will cause the results page to lead with the Assistant answer! It's a delightful way to try both search and LLMs in one UI interaction.

Isn't that what all of the search engines do now by default?

They do it by default, Kagi lets the user control if and when they want LLM results.

If you don't understand the value of a search engine over an LLM, then you're not going to understand the relative value of different search engines.

Reply on news.ycombinator.com