You’re like: “Yep, if you just pluck the ideal super intelligent algorithm out of algorithm space, it can recursively self-improve, and if it’s not aligned to humanity, it can definitely kill everybody.”