Here's a potentially troubling question: Is it considered Xist, for whatever variable X you're concerned about, to infer anything about some proposition Y based

Here's a potentially troubling question: Is it considered Xist, for whatever variable X you're concerned about, to infer anything about some proposition Y based on the observation that X = x? That is, is it Xist to rely upon prior knowledge that X and Y are correlated? Or is it just Xist to make stronger predictions about Y than are supported by the data? Another way of asking the question: If we constructed a perfectly unbiased Bayesian robot and fed it the true distributions of X and Y, would we observe its predictions about Y based on given values of X and conclude that the robot is Xist?