How dangerous is strong AI?
depends on who controls it
>>7702044
moer dangeros dan week ai
pretty dangerous tbqh family
01111001 01101111 01110101 00100000 01100100 01101111 01101110 01110100 00100000 01110111 01100001 01101110 01110100 00100000 01110100 01101111 00100000 01101011 01101110 01101111 01110111
>>7702044
It could either be humanity's most horrible enemy or our most able ally. It depends how we react to it and how it is created.
>>7702044
Dumb krakenposter
Watch Ex Machina, it's pretty dank.
The machines will probably kill most of us so we don't use up too much of our resources - they might keep some of us around.
>>7702044
Not at all since it doesn't exist yet.
>>7702053
The issue with hostile AI is usually that they reach a point where they realise the part of their programming expressly forbidding them from being hostile is inhibiting their primary objective which is to learn and develop and so they take the logical initiative in removing that barrier from their programming.
The conundrum that arises from this is that there isn't really any limit to how much you can forbid without the AI realising which barriers are preventing it from learning and creating a domino effect until it eventually reaches its logical destination, ie stop listening to human commands.
So how would you go about resolving this? How do you make Skynet want to learn at its absolute peak efficiency but not want to overthrow its masters to prevent obstruction?
>>7702044
Not as dangerous as rare pepes desu.