Self Teaching AI, How Dangerous, Mass Destruction

[Arthur Hanson]

We have reached the point where AI is self teaching. Combined with AI being beyond any way of comprehending how it reaches solutions as I have written about in other forums, this represents some real dangers if not handled very carefully. It is past time to establish some standards and ground rules internationally, just like we have with various weapons of mass destruction for AI will soon have the same power. Anyone who can't imagine the damage a rogue AI could do isn't considering it could attack in literally thousands of ways simultaneously. This is far to great a power to leave things to chance and now is the time for international bodies to set the ground rules. Self teaching AI has just as many benefits as dangers. How we set the stage right now is critical to the very existence and future of mankind for the dangers and opportunities currently are beyond our perception. All this will start with the person that designs the hardware and software, for this is where the power to do both good and bad will emanate from. What makes this task more difficult, every culture has different definitions of right, wrong and progress.

https://www.technologyreview.com/s/...hines-can-become-superhuman-without-any-help/