"Lacking certainty" seems to me to be very different from "age of AI" (unless you mean that AI gives confident answers of uncertain truthfulness).

There are two problems that you should address that I see.

First, how do you teach them not to believe everything an AI says? Well, how do you teach them not to believe everything a politician says? An advertiser? A peer? Do that with AI, too. It's not a new problem.

Second, how do you keep them from being emotionally attached to an AI? Give them hugs. Give them eye contact. Give them genuine attention from an actual human - lots of it - and pray that it's enough.