Even as it is now, whenever I use AI to dig down into some details about - say - vaccine safety, the starting point is a blanket "safe and effective" with study links. When I ask it to evaluate the studies - according to weaknesses that I happen to know about - it starts to say things like "yes, there are in fact issues here" and to give me better information, and by the end of the exchange, I'm getting something closer to the truth. But you have to know what to ask and where the study weaknesses are.
If a normie asks if vaccines are risky, or if MK Ultra could still be going on, or if OK City bombing is questionable... the answer is always going to be a definitive no, and most people don't know enough or care enough to dig deeper.
With explicit state control... yikes, it just gets worse.