Superintelligence is limited by the speed of light in this universe. It’s probably possible to create/evolve something significantly smarter than a human, but probably not 10,000 times smarter. Maybe not even 10x smarter. Not on a human timescale, anyway (I mean operating mode).

It’s possible then to have a superintelligence that is slow but mostly irrelevant to humans, or have a very high intelligence that is relevant to humans (and still dangerous, perhaps) but is not insurmountable to comprehension at some level. And no “foom.”

Unfortunately, the AI “debate” is full of utter morons on both sides, so you must think about it completely on your own, really, or you’ll never understand anything.