I understand why they are doing this. My Google search usage is easily down 50%+. I doubt I am unique here.
While there are times where I want pure search (Kagi, Old Google) I mostly use LLMs to search now and have them provide me links for source data.
When I do use LLMs as a search engine I always want it integrated into my AI workflows with access to tools and scripts etc. I never want to have a conversation with a website that is geared towards advertising me products.
I think LLMs are good answer engines, but terrible search engines. For example, if I just want the answer to "How do I foo this bar with a thingamajig", or "what kind of foo exists for bar", LLMs are 100% better because they'll give me an answer without trying to sell me their thing or pump me with ads. It also lets me go more in depth than the very surface level seo spam sites that appear when you do a search like this. On the other hand, search engines should be more like "<thing>s released 1908" or "<topic> and give you results talking about what you searched. If I search for C algorithm design, I don't want to learn what C is, what an algorithm is, and various other seo garbage. I want to learn about C algorithms and their design. If I search for influential books from 1908, I don't want "top 10 classics from the 1900s" or "hidden gems of 1908".
Currently, search engines are pretty bad at the second one because people try to use them as the first one
> I never want to have a conversation with a website that is geared towards advertising me products.
yeah man good thing LLMs are structurally incapable of being incentivized to sell you a product or render referral links, this is surely future-proof
It's tough for me to square the two things happening simultaneously in AI right now:
1. LLM Model providers are starting to charge real costs to users, revealing that AI usage is much more expensive than the subsidized rates we've been seeing for years.
2. Google is now using an LLM to answer every single google search that happens, for which Google bears the entire cost.
> My Google search usage is easily down 50%+. I doubt I am unique here.
The question though: Why is that?
Is your Google search usage down because LLMs are "so much better"? Or because Google actively chose to destroy the quality of their search results to juice advertising revenue, and appears to continue to do so to juice AI adoption?
> and have them provide me links for source data.
And therein lies the answer: You don't care about the LLM, you're just using the LLM as a means to get the good links.
Maybe I'm weird, but I find AI incredibly useful.
I've barely used Google for over 2 years.
I barely driven myself in a year.
I haven't written code in 6 months.
About 90% of my searches are straightforward enough that an LLM wouldn't bring any added value (and all of them happen straight from the browser address bar, either to Google or Wikipedia at a more or less 50:50 ratio). And for the rest, yeah, I just use Claude or whatever directly.
"I never want to have a conversation with a website that is geared towards advertising me products."
Since this is how Google makes all their money, why are they killing it off? Do they think people will eventually pay for LLM search? Do they plan to stuff the results with ads, not even sharing the ad revenue with the content sources?
That’s the thing… sure, this new search will be useful at times.
But I still want to also be able to do my normal, old school searching.
> I never want to have a conversation with a website that is geared towards advertising me products.
The advertisements fed the content, which fed the AI, which in turn feeds your AI workflows. AI is still not trusted unless it's output is grounded with sources.
I use claude/gemini as my homepage now (I have to keep switching as these companies make "updates" that periodically render their models useless). Even if I want to search for simple things, I would rather have an LLM wade through the result and extract just the information I asked for. SEO, and now mountains of slop content have made this necessary. Only a matter of time before the SEO industry in large figures out how to game LLMs too, making them equally useless.
I already saw a article recently about how to set up a business domain which can reliably show up in a search result and dump overly positive reviews into anyone's context.
For me down 99%
We all know Google search has been broken for a long, long time. SEO trash will fill up your first page with results from trash content generation sites that repeat the same thing, usually flat out wrong. Actual meaningful results are buried deep, if Google will even let out of the "In order to show you the most relevant results" hell hole.
My experience with AI searches is that they'll still be wrong a lot of times, but it will condense/flatten the content generating trash sites and give me alternatives from these deeper results. What I'm looking for is usally in there.
I'm not at all in the same boat as you; I do not and likely will never primarily rely on LLMs for information. But it's fascinating to hear that even folks who do don't find this approach useful.