Refutation of The Lebowski Theorem of Artificial Superintelligence

Let me start with a thought experiment put forward by philosopher Nick Bostrom. Imagine an Artificial Intelligence; call it “The Paperclip Maximizer”, or PM for short. PM has one goal: maximize the number of paperclips in the universe. In more theoretical terms, PM has a reward function that gives it higher rewards for higher numbers of paperclips in the universe. It also has a certain intelligence level, where intelligence is measured by PM’s ability to maximize its reward function. It is quite possible that PM would choose to increase its…