We can't say with certainty whether AI will ever pose an existential risk to humans given an infinite timeline. But what we *can* say with absolute certainty is that there is no direct path from current technology to a "hostile AI takeover" scenario -- whatsoever. Worrying about Skynet today is as if scholars from the 1500s said "gunpowder weapons have been making exponentially fast progress recently, so we may soon have bombs that could annihilate an entire province". It may well come to pass *eventually*, but that technology will have nothing in common with 1500s technology. What's more, when it comes to pass, any preparation effort you made in the 1500s (writing treatises and rulebooks?) will have been laughably worthless and off-topic.