I work at a firm that manufactures and develops discretionary AI. In my opinion, general intelligence is a myth because computers inherently do not contain purpose. We are successful because we have experienced traders that program which sectors our AI should study so that it can focus and outperform with high depth. Letting a supercomputer think autonomously is like telling a self driving Tesla to drive without a destination. It will wander aimlessly and produce nothing until it runs out of power or breaks down.