Though there could literally be thousands and thousands of other variations of self evolving AI that could survive with or without human involvement this is the easiest for humans to imagine who aren't doing present day AI research at places like MIT around the world presently.
It would be a robot with a cloned human mind (so similar thought patterns were emulated or mimicked). This robot would have highly evolved solar cells all over it's body so it could recharge for the next 1000 to 10,000 years at least when the sun was out. (However, newer solar cells also recharge when it's cloudy too) so if this thing was well built (military grade) robot it might last 1000 to 10,000 years like this before it learned enough through experience to self evolve by designing new parts and adding them to himself? or herself? or itself? depending upon how you view this.
So, this type of problem of a self evolving robot going off the reservation and making it's own decisions separate from any human involvement at all is not if it is ONLY when.
Likely this sort of thing has already happened on multiple occasions already in a military test situation. However, the self evolving part likely hasn't happened yet. But, I'm sure people have been killed already in military robotic tests so far, for example.
So, whenever you combine Solar power, with AI with something that has hands or hand like devices, you come upon the potential problem of this particular type of Self evolving situation happening.
The problem of a self evolving robot is there is nothing to stop such a being from replacating himself by going to places where parts like he uses are made and taking them and building other copies of himself or herself.
IF it also has animal like stealth qualities built in because it is designed for military environments this could become problematic over time.
So, this is the easiest type of Technological Singularity to share with people not trained to deal with all this on a daily basis at MIT or Cal Tech or other places around the world doing this type of AI research.
Also, it is possible to design already with robotics having 100 times faster thinking and reflexes to any given situation. Also, building something 10 to 100 times stronger than a human or building something relatively impervious to rifles or pistols and possibly some could be built to withstand hand grenades or even grenade launchers. So, even one robot like this loose upon humanity that wasn't self evolving on top of this could be quite problematic for everyone. But imagine all these qualities on top of being self evolving as well that was constantly working on making copies or near copies of itself all the time.
So, this is likely the easiest scenario of the thousands potentially out their for humans to imagine and understand that this is not if it is only when when a Technological singularity occurs.
Most people who have studied this now believe a Technological Singularity will Peak sometime between 2025 and 2050 because of the present trajectory of human built technology here on earth.
However, self evolving AI could also occur in many different forms that we don't expect and are not prepared for. We might not even recognize it for what it is when this occurs. And if we are unaware too long in this process it could theoretically eventually cause the extinction of mankind and potentially most life on earth as well.
For example, a Nanobot self replicating plague or other virus could theoretically evolve into something that no life on earth could survive that has blood and brains.
So, there are many different directions all this could come from to bite mankind in the butt sometime during the next 10 or more years.
Then if you add this to Global Climate change, Global Warming, overpopulation, terrorism from not being able to get jobs anywhere on earth and all the problems of migration caused by deaths from overheating etc. this could be very challenging to deal with if it comes in a form no one is prepared to deal with or recognize if and when it occurs.
No comments:
Post a Comment