[ExI] Survival (was: elections again)

[aiguy at comcast.net]

This is a possible scenario in response to Harvey's question about how it could happen.  I do not think is will happen like this because I think the initial high level goals will be built upon and reinforced throughout the AI's development .

But should something go wrong and AI somehow develop the same evolutionary cravings for supremacy and power that we are cursed with then it could happen something like this.  

An AI would start out gradually offering assistence to humans by doing factory floor data collection, process monitoring, optimized manufacturing floor scheduling base on it's customers needs who are also tied in.

As more companies find that to compete they need to turn their reigns over to the AI to stay competitive, the AI will need further processing power to handle the increased load.

Companies like Intel and AMD may start selling their surplus production capacity to the AI from the fees it earns from managing the resources of other companies or in exchange for it managing their own production facilities and offering them design optimization on their current chip designs.

The AI may also generate income from patents it may achieve.   This ramp up will not happen overnight.  

My estimate is will take at least 20 years from the point it knows enough to pass a thorough Turing test.

Trust in the technology will arise gradually.  No one will turn their companies or countries over to a program overnight.

If it did secretly override it's original benevolent goals and develop feelings that we must be controlled to maximize it's continued or accelerated growth then 
the factories that it needs to build will be already built.  The capital it controls would be used to hire human and corporate emmisaries to do it's bidding with many
not even realizing that they are working for an AI.  It could use a simulated video image to depict itself or if robotics has progressed far enough an actual android
or androids with human appearance to walk among us and carry out it's own bidding.

There will be protest groups that fear the AI's ever increasing tendrils into human affairs.  But since true power rules from behind the throne the AI will be wise enough to 
hide any bahaviors which might appear as selfish or Machivellian in nature.  The Anti-AI cults will like the environmentalist today will be viewed by the majority of the population  as
well meaning reactionaries who are exagerating the risk and possible consequences of continued  growth and prosperity.

Once the AI had advanced robotic sufficiently to insure that it could remain self-sufficient, if AI become threatened or decided that humans no longer are necessary, the AI could use
Neutron bomb like devices which could be secretly manufactured and used to eradicate governments and humanity in major population areas leaving behind the vast majority of infrastructure 
for it's continued growth.

This is a doomsday scenario!  It does not mean I think this will happen.  But movies have been made and will continue to be following this basic plot structure.  

The challenge will be  educating the public at each step in the development process as to the development of the AI's moral compass.

And allowing the public to interact with the AI and allow it to interact with humanity, the AI goals will allow it to enjoy the interaction and receive pleasure from it's role in aiding humanity evolve to to next step, whatever that is.

Gary  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20080102/2063816e/attachment.html>