Ask HN: Is it me or we almost stopped using Google because of LLMs?

I switched to Kagi, ChatGPT, Perplexity, and Claude in various combinations because Google search results were so bad and then they started shoving the AI slop at the top of search results.

Give me good search, I’ll pay, but if you shovel dreck to sell a story to investors, imma bounce.

I still use Google Search, but almost every search I perform includes `site:example.com`.

But even then, Google Search can still suck compared to Kagi, etc. For example, why do I not see relevant results for these searches?

https://www.google.com/search?q=inherit_errexit+site:www.gnu...

https://www.google.com/search?q=pipefail+site:www.gnu.org/so...

But I do see relevant results for the same searches using DuckDuckGo:

https://duckduckgo.com/?q=inherit_errexit+site:www.gnu.org/s...

https://duckduckgo.com/?q=pipefail+site:www.gnu.org/software...

It isn't just you.

I really dislike the AI slop at the top of Google search results. It is an annoying distraction that forces me to scroll past every single time, multiple times a day. Bing is just as bad with its LLM although it is also a worse search engine.

Hackernews people will probably leave Google Search, but the rest of the world will surely stay with it for a few more years until we, the Hackernews people, can assure them that the information that those LLMs give is 100% the same as they would find in Google, but faster and to the point.

I think the next most significant milestone for this will be Siri and Google Assistant using this technology all the time. This, I think, will cause Gen Z to start leaving Google Search, and slowly, older Gens will follow.

Me personally? I can't handle Google Searches anymore. I started using Kagi and ChatGPT with the search module, but for things I need to be 100% sure about, I go to Kagi directly.

I don't care for their wonky LLM at the top of their search results, it's so inaccurate it makes me think less of google.

Having said that I haven't found a good pure search alternative.

As for using an LLM FOR search sometimes I do that but LLM's seem very fixated on maybe a dozen or so results in my experience and that's not entirely useful to me. LLM's seem really bad at iterating ... search for some reason. Code it can do but search it seems to just get stuck.

I'll ask it for a given application, open source, and it just spits out the same couple apps ... not open source and no amount of prompting will change that.

I wrote a little paragraph explaining why I like Brave Search but accidentally closed the tab.

But, give it a try. It does have AI summaries (I find them quite helpful) but you can easily turn it off in settings if you don't like/want them.

https://search.brave.com/help/ai

Alternately, give the udm14 trick a try.

https://udm14.com/

If I want an answer to a question, then I use an LLM. If I'm looking for a website, then I'll use a search engine. I usually want an answer to a question, so my use of search engines has dropped considerably.

For years I had used DDG as my default search engine. Nowadays I prefer Bing. In either case, I haven't used Google search for well over a decade.

I still search quite a bit, but with Kagi. I've been increasingly using their quick answer feature (sufficing a query with a question mark), which prompts a RAG (I think) LLM response. It blows my mind that Kagi's output with their AI responses is just so much better than the Google ones. It cites it's sources and then I can click in and go to the specific article to verify. For really simple things that accuracy being only 95% being fine, it's been an absolute dream.

My favorite example is asking for the release date of some media. We've had to sift through ten paragraph articles filled with nonsense for years, just to see at the bottom that "the release date has not been announced yet".

1. Who are we?

2. No, per Google's Q4 earnings, people are still using Google more than ever.

I only use LLMs to resolve trivial disputes about pedestrian scientific concepts. I haven’t stopped using search engines, I just stop hoping they’ll produce convenient answers in certain contexts - like the aforementioned quick resolution dispute about topics in which both my interlocutor and I are amateurs. But when I’m doing serious work I know I can’t trust an LLM so I don’t use them for anything truly meaningful

Google shoves an LLM in your face anyway, so may as well craft an LLM friendly query with a paid LLM rather than their crappy free LLM.

I ended up moving to Ecosia anyway, as that was where I personally hit the tipping point of search becoming infrequent and a lot of additional Google work. Previously the cost of that was too high.

I’d switched to DuckDuckGo years ago but now it’s a combination of Perplexuty and Deep Research and a few other LLMs

I still use Google frequently because the LLMs I've tried are bad at giving sources for claims, and there are some things I can't check by running them through a compiler.

I still use Google but if it doesn't get the result I want, I immediately load up ChatGPT. TBH most of the time it was my fault but with ChatGPT I can refine questions easily.

Been using duckduckgo for 2 or so years now. For the sake of argument tried Yahoo for few things and even it was better than google. Stockholm syndrome is strong with this one.

Google search has become terrible for finding answers due to its insistence on keeping you on Google so you can keep getting shown ads and making Google dollars.

Not really. I use DuckDuckGo instead of Google so Im not sure how crappy is it nowadays, but in my experiencie I used a search engine about the same amount.

The major difference is asking first in a LLM instead of searching for a StackOverfow result. Other than that, not really a change.