>>12652958>>12653047>>12655901As someone who studied/studies the human brain academically, unless you are a dualist (or have some other interpretation of """"consciousness"""" and """"human thought""""), you all seem to not understand that (unless something physically stops our computers from becoming powerful enough) we will eventually be able to simulate a human brain. And that simulation will have the same amount of consciousness and originality as any other human.
Firstly, the simulations we have right now really don't work much like human brains do. How does an AI work at the moment? As far as I'm aware, you produce a series of layers of nodes, one input layer, one output layer, and many hidden layers. There is a general directionality, input > hidden layer 1 > hidden layer 2 > ... > hidden layer n > output layer.
Maybe later hidden layers are able to impact earlier hidden layers.
Then you train these layers on a sample set of data to minimize inaccuracy.
So, in comparison to a human brain, we have software that tries to approximate the human brain, but fails miserably. The human brain has a general layout that develops as a function of evolution, by genes regulating the development and positions of cells, and their developmental fate by chemical gradients. You could call this an evolutionary blueprint.
Researchers are still in the infancy of producing artificial neural network blueprints.
Furthermore, human brains form new neuronal connections somewhat spontaneously, in real time, as a product of hebbian learning. Computer AI don't at all seem to work this way.
Lastly, we don't at all train AI on a dataset comparable to the dataset that humans went through when we evolved. You can see that training computers on games of hidden information may lead them closer to a "theory of mind" (as it is called in psychmemeology), but it hasn't yet approached anything human like. Arguably because evolution has the key trait of selecting for survival.
