Concerned that AI and self-improving machines could accelerate and go astray, potentially undermining humans, Elon Musk has started a company that can allow AI to interface with (and remain controlled by) human brains.