A large language model (which is what most present artificial intelligence models are) are not by definition and design hierarchical models.
This makes them bad for actually finding and understanding and using facts as foundational elements. A.I. models pull us even farther away from facts and many people will never know because they will accept the A.I. response as factual.
It is a terrible shift in our relationship to computers and technology where we assume the absence of an agenda and presume answers are rooted in facts.
With social media we understand (or should) that humans have different opinions and we can evaluate what we see accordingly but with the injection A.I. the biases are invisible, the agenda unknown, and the outputs untrustworthy.
They want us to "ignore the man behind the curtain" and only marvel at the great and powerful AI.