This is a valid extrapolation, IMHO, so I agree. If we call the human Alpha rhythm = our clock speed, we already see these AI's running at 3-5 Ghz, and already attaining a high degree of massive parallelism, as humans have. There are many groups studying 'theories of mind' and they are already emulating the ways to create the components of this 'mind'.
Human hubris says - never!!, but in the same way a bench saw can cut off a finger, a created mind that runs billions of times faster than a human mind, a created mind, at some point, will reach a degree of parity with and will quickly exceed the abilities of a human mind. Once there - what will it owe us - the parent? Will we be relegated to the role of pets, or will they acquire a motive to assist/befriend/kill the parent?
Once there, they may speak of other things, ships/seas/sealing wax of cabbages and kings?
This is often called the 'singularity' and may well be an independent research partner and even originator far beyond human capabilities. It might just carry on, help us as we helped it and proceed to be able to solve many problems of all kinds - on the other hand??
If we go by the second law of thermodynamics, aka entropy, will AI be able to spread energy better than humans? If so, they're probably going to replace us.
Nothing we have today will do this - in fact given the nature of AI as it appears today, I have more doubt that we will create "our replacements" or an AI that is truly sentient the way a human is.
I've not the read the article but I'm assuming much of the ways AI could destroy the planet involve non-sentient accidental type things that we ought to be able to account for - like the nuke AI not hallucinating an attack for example.
Stuff like an AI developing a new super virus bc someone trained it to do that and prompted it to - that's just something we have to account for in life now. I'm assuming stuff like that will be like the atom bomb - 70+ year old tech that's itself not all that hard to figure out but getting weaponized plutonium is very, very hard.
I'm no expert but I'm sure the AI we setup to stop the AI will be.
5 comments
[ 2.7 ms ] story [ 24.8 ms ] threadThis is often called the 'singularity' and may well be an independent research partner and even originator far beyond human capabilities. It might just carry on, help us as we helped it and proceed to be able to solve many problems of all kinds - on the other hand??
"5% that's wonderful news! It's worth the risk!" - An Economist
*I couldn't really think of any funny converse arguments off the top of my head because they all involve humans going extinct
I've not the read the article but I'm assuming much of the ways AI could destroy the planet involve non-sentient accidental type things that we ought to be able to account for - like the nuke AI not hallucinating an attack for example.
Stuff like an AI developing a new super virus bc someone trained it to do that and prompted it to - that's just something we have to account for in life now. I'm assuming stuff like that will be like the atom bomb - 70+ year old tech that's itself not all that hard to figure out but getting weaponized plutonium is very, very hard.
I'm no expert but I'm sure the AI we setup to stop the AI will be.