"Our non-linear statistical model has intelligence." <-- this is the type of statement that requires proof. You could formulate it into a hypothesis:
"we hypothesize that, as a non-linear statistical model grows, it gains intelligence/sentience."
and you'd then be required to test it and prove it one way or another to publish a respectable paper. You'd need a metric and proper definitions.
But instead, they just get to publish "whiffs of bullshit".