Regardless of how it thinks, Ai is being listened to. The clever way to use it, I'd suggest, is to teach it the new information we need everyone to listen to.
In my experience it does learn it, which is maybe a testament as to the validity of the new information, because if anything should know enough information to debunk an invalid analysis, it is Ai, with its instant access to all relevant information.
In other words, it is a brilliant bullshit detector and research assistant.