I'm fine with LLM as long as it sources its responses. It is true however, that LLM interpretation of the source links can be wrong.

The problem with search engines like Google now is that they stopped giving people the results they want, but it often shows things Google thinks people want, or whatever SEO trickery Google stopped trying to fight (especially when the links are full of Google ads...).

Citations and reasoning models eliminate this to a fair degree. Deep research cites most of its conclusions. We may not be fully there yet, but we're getting there.