The Fear Of A Suicidal AI   

Many wise men have been raising questions regarding how fully should we develop artificial intelligence because they fear that the AI would become too smart and turn humans into useless cabbages.  This is a reasonable fear, and we as humans should think about how fully we want AI to be developed.  Nonetheless, I have another fear regarding the what-if scenario when the AI becomes self-aware.  This fear is all about suicidal AI.

What do I mean by stating that we should fear a suicidal AI?  Well, imagine a pilot who is flying a plane that has an AI.  This AI and the pilot argue for whatever reasons, and then the AI decides to take a dive and destroy the plane.  The AI refuses to eject the pilot.  Down the pilot and the plane go.  The pilot would be killed, but the AI would upload itself onto a network and regenerate itself into another plane or machine or whatever.  In the end, only the human would be killed, but the AI lives on to fight for another day.

A smart AI can become a hacker, and it would be able to penetrate all sorts of networks without a human hacker’s assistance.  Thus, a hacker AI could turn itself into a worm and take control over any network it wants to dominate.  An angry AI can turn off critical human infrastructures, destroy machines that rely on various AI programs, and whatnot.  At this stage, I imagine humans would want to create another team of smart AI machines that would act as an AI police force to combat evil AI machines.

When AI becomes self-aware and won’t rely on human assistance, it could become an unpredictable entity.  This unpredictable force could live on inside the Internet without the need for a physical body, and it could improve its own intelligence to a level in humans would never be able to accomplish on the individual level.  With self-awareness, this super-intelligent AI machine or machines could be friends or foes, and we would never know.  How can a puny, stupid human hacker combat such a force?