There is an underlying issue with how AI presents sources in a way that doesn’t make it clear why those sources were selected or if they are the current source.

For example, one of the few times I tried AI the sources were older versions of documents not posted to the current website or in half of the cases contradicted the ai answer text. My favorite example is that it used a proposed law as a skurce for a change that never made it out of committee, but also included a link to the actual law that had the right information. A good search engine would just link directly to the law for the correct answer. If it was right a few times in a row I can see someone not checking sources after a while and that introduces a trusted vector for misinformation.

At least with a search engine I can generally tell if the site in the results is reliable, like a university or known organization.