There are so many mundane social problems with the technology here and now that don’t even remotely range into the territory of Skynet, Paperclip Maximizer/Grey Goo, Roko’s Basilisk or The Matrix. So many huge, conspicuous, world-changing events have to happen before any of those sci-fi situations are even close to plausible, yet those are the things getting the policy attention.
Baldur Bjarnason included a link to this article in a list of articles he recommended. Taylor breaks down all of the things that need to happen before AI can become sentient and destroy humanity or crush us under the weight of a bajillion paper clips. There are lots of reasons to be concerned about AI, but we’re a long way from fending off terminators.