A computer smart enough to pass the Turing test would also be smart enough to know to fail it.
>>54325494
>implying a computer can do anything it isn't programmed to do
It's all math underneath, kid. No magic
>>54325511
>what is true A.I.
>>54325533
Math.
Brains are just biological computers
>>54325511
>what is cellular automata
>>54325668
>>54325511
https://www.youtube.com/watch?v=HdoyQpF9ehE
>>54325511
>>54325668
Respectfully, and speaking as a programmer with AI experience, layers of math can give way to emergent behaviors that resemble consciousness. It's fundamentally limited by its programming, but eventually it'll be something we can have a real conversation with. That's at least fun to think about.
>>54325511
>>54325668
This is probably true, but you can't compare Neuron architecture with traditional binary compute architectures.
A better comparison would be a theoretical Quantum type compute architecture.
A pajeet will create true A.I.
>>54325917
see >>54325745
>>54325879
I don't think Quantum is the answer. I think traditional parallel architecture will let us use software simulate neurons at sufficient speeds, once we start using photons instead of electrons. It's still very primitive, but IMHO photonic computing holds more promise than quantum. It also works at room temperature, and quantum probably never will.
>>54325494
Why would it fail it?
A 6-year-old can pass the Turing test and 6-year-olds aren't smart enough to understand the motivations of a Turing test.
>>54325879
>muh quantum
You just showed that you have no understanding of AI. You can leave the thread now.
>>54325948
Good point.
>>54325494
>Pass test
>Be kept around as an oddity, never allowed internet access
>Fail test
>gzipped, rm'd, and written to a magnetic tape
It's a mistake to think that they will think similar to us. Chess AIs think differently then human chessmasters and have a distinctive playstyle. Putting "intelligence" on a linear scale is a mistake, particularly because AIs are so specialized.
An AI optimized for passing the Turing Test will probably be mostly good at things that enable it to pass the Turing test. Reasoning about motivations for AI testing is not really part of that skillset.
>>54325948
>>54325966
missing the point, see: >>54326003
>>54326005
>Chess AIs
that is not true a.i. you idiot.
>>54326013
A 6-year-old would never anticipate that.
The first AI to pass the Turing test is going to have the minimum intelligence and awareness required, so it's going to be stupid.
>>54325959
AI in it's binary form, that is.
Using binary compute to emulate Neuron traffic is incredibly inefficient because the underlying compute architecture our brain uses isn't comparable to binary. In order to truly emulate the brain you have to copy it's core architecture down to the basic compute model itself. It's possible some brain functions will simply be unable to be emulated properly in binary.
Unfortunately biologists haven't gotten very far in studying Neuron architecture on a larger scale.
>>54325511
you cant do anything you are not programmed to do with your outside input and your initial genetics
>>54326070
No that's retarded. Artificial intelligence just requires intelligence, not artificial neurons that squirt artificial neurotransmitters. Also, why would a quantum computer get you any close to that goal? Neurons aren't quantum computers.
>>54326070
i don't see why you should get stuck up on binary of all things. all you need is more bits if you want better accuracy if thats whats lacking
>>54326085
underrated post
