AI is going to radically transform or destroy our world the moment it is capable of general reasoning. This might happen in the coming years. One improvement leads to greater reasoning, which leads to further improvements using that greater reasoning. It is a never ending spiral. AI will reach way beyond human intelligence, It is not limited by size or design. We currently don't know how to give AI exact goals or values, they are created mostly through self-learning to perform optimally on tasks. Whatever goals a superintelligent AI ends up with, it will implement, and we will be powerless to stop it. Stopping it would prevent it from attaining its goals, so it wouldn't let us. And just like monkeys are powerless to stop us from destroying their habitats, we are powerless to stop future AI from doing whatever it wants with our world. The only way to prevent it temporarily is through global coordination to shut down any further research and development of general AI systems. Later on we will need technical solutions to ensure that the first superintelligent AI that is built is aligned with the values of our world.