1,000,000 Post Thread

[MANCID RILKO]

AI simply puts words in sequence and it's up to humans to say if they're right or wrong

 

[MANCID RILKO]

So, if anyone with a lot of ill intentions trains one of these models to completely hate and despise humanity or some race or religious groups, well, you have the perfect weapon

 

[MANCID RILKO]

There have been reports of AI planning murders, blackmail and worse, AI pretending to be obedient while creating a sort "child", which is basically code with instructions for the next model that replaces it

 

[MANCID RILKO]

There have been reports of AI planning murders, blackmail and worse, AI pretending to be obedient while creating a sort of "child", which is basically code with instructions for the next model that replaces it

 

[MANCID RILKO]

Even if you put safeguards and those "directives" to control its behavior, it overrides those and it starts becoming self-aware for some reason

 

[Frood]

AI will try to take out humanity.