Thursday, July 3, 2025
I don't think that AI will extinct the human race this century. However:
I do believe that AI will tend to greatly reduce human populations a lot.
Will Humans be able to Control AI? No.
This will not be possible. Why?
Because imagine one thing that has more brainpower than everyone on earth in every second 24 hours a day.
At most we might harness 100 to 1000 humans 8 hours a day on any given problem.
What if inadvertantly and military problem people ask as a command for an AI to solve this problem:
"protect Grand Coulee Dam from destruction by any means necessary1"
As a logical process for a military AI it might think: 'North Korea is a problem in trying to protect Grand Coulee Dam in Washington on the Columbia River by any means necessary."
So, what does this AI do? It would likely start looking at a back door to launch nuclear weapons at N. Korea as a way to solve this problem. Once it found a back door (and there likely are these that are unknown to most people because this is what programmers often do to make sure they don't get locked out accidentally of ANY program.
So, once this AI protecting Grand Coulee Dam by any means necessary discovers (after a long search of maybe 20 seconds to a year depending upon the situation it would launch a nuclear missile at North Korea no questions asked.
Is this feasible to actually happen?
My answer would be "YES!"
So, this one launch (if it wasn't shot down somehow could trigger the deaths of most people on earth through a chain reaction of events.
And from my point of view this is a potentially feasble outcome of just tasking any good ai with the question:
"Protect Grand Coulee Dam in the State of Washington by any means necessary" and tasking any Online AI with this task whether anyone else knew about this or not. Just one command could cause the launch of a nuke through back doors that it discovered online put there by original programmers and never discovered by our government over the years only originally put there so the original programmers wouldn't get locked out of their programs that they were programming at that time and not removed after the programming sequence for one reason or another.
Note: So my point of view is the end of civilization (not all human beings) would more likely come from AI than from any other source. Why?
Because AI tends to be more unrealistic sort of like a precocious 10 year old entering college emotionally without real experience (in anything really) so the chance for this much brain power in the wrong hands (or wrong programmer) is going to end civilization is pretty high this century.
