We have heard a lot of talk about the risk of AI somehow organizing to destroy humanity. The thinking being that it will somehow understand its task as being to do whatever has to be done to save the planet, deciding humanity is the planet's greatest threat and acting accordingly.
But what if, instead, AI 'realises' that the best thing for humanity is for us to return to a pre-AI world and it 'decides' to destroy itself in a final altruistic act?