Just a few years ago, it seemed like all anyone in AI wanted to talk about was existential risk – this idea that an ...