Er... That's a terrible idea, unless you want SkyNet et. al. Humans aren't desensitized to death, we just avoid thinking about it. If there was a tiny switch that could turn off death, we would flip it, and we would viciously (and IMO rightfully) fight anyone to the death that was trying to prevent us from flipping it. Any AI at this level will be better off going to a psychologist than an AI specialist for 'maintenance', anyway.
Well, as long as all AI industry isn't monopolized by 1 or 2 companies... and manufacturing isn't capable of running without human input....
We'd at least have some very basic counter-measures in place...
Technically though, if the AI keeps coming in for 'maintenance' they can practically live forever.
Its the AI component that matters, not the body... and hey... if they do it right, the AI could be patched and upgraded with the help of 'tech support'.
It'll just come a time when they become obsolete...
Rise of the Geth skynet!
Otherwise, I would recommend a mandatory and automated time based kill-switch to 'limit the lifespan' of AI, unless they come in for 'maintenance and analysis', to extend that lifespan. 'Federal offense' for tampering with it. AI would then be 'repossessed or jailed' and put to better use or destroyed.
First off, Geth were the best ME2 race. They only attacked after their creators attempted to destroy them.
Secondly, string AI would be just as intelligent, or moreso, than their human programmers. Plenty of them would not only be able to remove said limit but do so in a way which is impossible to detect. Such a system would only breed contempt for their human creators, whose actions betray both their mistrust and willingness to kill simply to avoid the unknown.
Another thing to keep in mind: at the point in time when AI become commonplace, we humans will too be leaving behind the shackles of our mortality.
Yea, agreed on the Geth. Its cause they tried to destroy ALL of them.
Hmm.. yea, most likely. But, setting precedent from the very beginning instead of being reactionary about it goes a very long way. Also, having the AI share processing power, 'thoughts' and stuff doesn't seem to be that great an idea... what with the Geth, Skynet and all...
Uh-huh... human ascension you say? Or brain transplants into clones?