Even worse: this kind of "AI" can't be applied to systems we *don't understand*. We can observe far more causal relationships - in which A reliably causes B - in the universe than we can explain. There are lots of examples of us operationalizing these observations without understanding them.