The problem with this approach is most interviewers are not willing to be convinced that they are wrong. I've had a high-level engineer from Apple argue fervently in an interview that the Boyer-Moore string search algorithm 'can't possibly work' and a high-level engineer from Google argue fervently that linked lists are superior to arrays (despite me clearly explaining the problems with pointer-chasing and locality).

Some interviewers just come in with their own preconceptions and outdated/incomplete information, and I very rarely encounter interviewers that actually appreciate learning something new from a candidate. They do exist, but in practice you're better off just shutting up and waiting for them to move on. Pretending to agree doesn't help anyone, of course.