Say we wanted to replace our government with an artificial intelligence as a sort of enlightened despot.
What might the software architecture look like?
Is it within the realm of our current knowledge of algorithms and data structures, albeit exceedingly complex?
Would it ever be completely safe (no doomsday scenarios)?
Lets discuss.
that's a fucking horrible idea
Nice try skynet.
>>58890901
>>58890910
I'm not so sure. In order for the AI to kill us it would need:
A) Direct control over weapons of mass destruction
B) Direct control over machines to keep itself running
So long as we don't implement a network connection between the AI and the powerplants/mines/factories that keep it running, it can't kill us without dooming itself.
>>58890887
Nice try La li lu le lo
P A T R I O T S
A
T
R
I
O
T
S
>>58890887
plz be my ai president
>>58890887
it's probably better than the current batch of idiots.
>>58890887
"AI" can handle things like the flow of traffic. It can handle accounting, high frequency trading, it could draft a balanced budget. It can't handle philosophical issues like individual liberty, and at what point a government can encroach upon them.
You can't let software decide what constitutes free speech, if a gun control law is constitutional, or if spying violates the 4th amendment.
>>58891166
>You can't let software decide what constitutes free speech, if a gun control law is constitutional, or if spying violates the 4th amendment.
yet
>>58891166
You may be right about philosophical issues like individual liberty, but I would disagree about topics concerning constitutional law. Some primitive legal expert systems have been developed that, given enough work, could eventually be able to examine constitutionality as well as any human lawyer.
I'M ELECTED ELECTRIC SPY
I'M PROTECTED ELEEEEECTRIC EEEEEEEEYEE
Something like that?
M A G I
A
G
I
>>58890887
It will look like ryzen and be maintained by cyborg keller
>>58890887
>AI in government
Have we not learned anything?