Yes, the truly interesting models are hard to analyze but, to paraphrase, if intelligence was simple enough that we could understand it, we would be so simple we couldn't.