The real problem here is assuming this takes off what incentives will anyone have to provide the information to feed the beast?
An llm rephrashing / regurgitating other websites is imo different, because you loose the direct connection to the original source. Even if llms give sources they also directly give you a plausible (but unreliable) answer to your question. They are right often enough that you get lulled in to the false sense of security of not needing to read the original sites. I'd much prefer them to just give a clean list of sources like early google, but then why would you need an llm.
It's a pity that probably the main reason you'll need an llm to find anything on the web is to weed out all the llm-generated low quality garbage.
I disagree, Alta Vista had excellent an excellent search UI with Boolean logic. Google discarded that because it thought it 'knew better' in terms of Page Rank.
A-V could be fine-tuned to find a page with exactly the search terms, Google just gave fuzzy approximations from a very large search set.
This is precisely what all the extreme LLM haters have memory holed.
The end of search traffic will kill all but the largest sites, and prevent countless new ones from being developed or getting traction. Given how global trends are going I expect the remaining sites to be increasingly monitored and censored/biased. I'm not looking forward to a world where social media means talking to some bots tuned specifically to addict you, and don't know too many people who are. Although big tech executives certainly seem to be in the latter group.
Did AltaVista get replaced by the owner of the site to justify a giant investment?
Now, the spam is back and it’s coming from Google itself.
Of course, even Google the search engine has gotten worse at surfacing interesting websites. First came the SEO spam websites, now the slop websites.
I'm glad that alternatives like Kagi exist.