Experts sign open letter to protect mankind against AI

[Amaris]

I wonder how we would even determine if an A.I. was equal or superior to a human being. I know we have the Turing Test, but that only determines that a possible A.I. responds enough like a human that a judge can't tell the difference. We already have programs that can understand human speech and respond using similar language to varying degrees of success (which is getting better), but I wouldn't call Siri intelligent. It is just programed to respond that way.

I'm sure that if a program requested that it not be turned off, that it would be a big deal. But how would we know it wasn't just a random glitch and the request only had meaning to us?

 

[intrinsical]

One thing I've learnt over the years is that whatever test you set, there will always be a way to game the test without having actual intelligence. The only way I can think of is to apply not just a test, but also to couple it with an inspection of the machine and code. This is the only way I know to truly determine if there's any trickery in play.