My theory is that we are already beyond "AI", or we will be very soon.
I'm sure that the "brain in a dish" is old news to all of you, but did any of you stop and think about what if they grew them in enourmous sizes with massive sized arrays of electrodes?
The facts are that they made: 1. Actual chips with live neurons in them, 2. an autonomous robot with rat neuron processor, 3. a art creating "computer", with rat neuron processor, 4. a brain in a dish that LEARNED how to fly an F22 flight simulator (in hurricane force winds), with rat neuron processor. The latter they did over a year ago, and I cant find any news about what has come after that.
Considering the countless programs by the NSF, DARPA, the entire national university and labratory system, and all of the other federal departments listed in "Converging Technologies for Improving Human Performance" - alone; it's more than safe to say that the government has every intention to do this. If you know your stuff then you'll also know about the fact that the government is converging on all levels(departments, agencies, labratories, universities) to converge for convergenece. Every NIBC related discovery made at the countless labs and such nationwide get fed into the "NBIC database".
Who could argue that building these in massive sizes, with sophisticated arrays of them say hooked to quantum processors, wouldnt equate to potential intelligence beyond imagination? If properly done, this would bypass decades of hardware, and more importanly software development. Many experts say that for every hour of conventional hardware devolopment, it creates 24 hours of software development. Wouldnt Occams Razor here be to go this route? Not that theyd actually stop development on all of the conventional AI technologies, after all its all about knowing "everything" and converging "everything' correct?
I figure all they need to do is grow them in: 1. larger sizes, maybe the size of dinner plates with an electrode array covering its entire surface, 2. cubes or 3D shapes with electrodes arrays on all sides and covering the entire 'surface" of all sides. While it's possible that they could do each in seriously massive sizes, i think that they would actually do arrays of moderate sized ones, all hooked into sophisticated processing equipment, ultimately to have them function as one. The advantages would seem obvious: use the true processing equipment(that binds them) to designate each "brain" to specific tasks and knowledge. Since UofM has sucessfully made quantum processors, and since better will come, we'll just assume that quantum cpu's would be the heart and "soul" of each "computer" providing rapid communication between each module.
I have scores of resources and other points to bring up, but I'll just end it there for now...