Google's AI overviews gives you sources. The sources don't always say what Google AI thinks it says. Or, Google will mix up two sources that are talking about totally different things, assume they're the same thing, and then show you nonsense generated from the mix.

I don't immediately assume that Perplexity will be any better off. Citing sources is great, but I'd rather just read the sources myself rather than assuming that the AI did anywhere a good job of actually summarizing them properly. At that point, what does the AI actually usefully bring to the table?