Microsoft's president says he doesn't think artificial intelligence poses an immediate threat to humanity's existence, but "safety brakes" are needed to address AI's risks.