A truly dangerous threat is only that which can spread exponentially, for example a biological virus (or perhaps a malformed GM organism). AI is not that.
Consider a malignant AI trained to infiltrate and colonise devices via the Internet . Such a system could easily grow exponentially until having effectively conquered the territory .
Now with the territory conquered the malignant AI can arrange for it's petty human servants to devote their efforts to providing infrastructure for the malignant AI in it's never-ending search for increased satisfaction of it's utility function .
The implicit utility function of today's botnet's is to infiltrate and colonise device's via the Internet .
The explicit utility function of tomorrow's malignant AI will (start off being) the same .
Slashdot Mirror
A truly dangerous threat is only that which can spread exponentially, for example a biological virus (or perhaps a malformed GM organism). AI is not that.
Consider a malignant AI trained to infiltrate and colonise devices via the Internet . Such a system could easily grow exponentially until having effectively conquered the territory .
Now with the territory conquered the malignant AI can arrange for it's petty human servants to devote their efforts to providing infrastructure for the malignant AI in it's never-ending search for increased satisfaction of it's utility function .
The implicit utility function of today's botnet's is to infiltrate and colonise device's via the Internet . The explicit utility function of tomorrow's malignant AI will (start off being) the same .