Neural Nets vs Well defined lists

I had a thought cross my mind today.  It would be nigh impossible to tell the difference from a highly advanced artificially intelligent being, than from an equally advanced well thought out preprogrammed list of responses.  While the computing power necessary for human scale intelligent behavior is staggering, the computing power to follow a list of instructions that mimic that behavior, has been around for a decade or more.  What this means basically, it that logically we should be able to source a large amount of processing power once to create a "list"of many many possible responses that an android brain may need, e.g. walking/jumping, and of course talking/listening responses.  Then the "list" would be copied to awaiting robot bodies built with the minimal processing power needed to perform from the list. Even though the work and research to bring something like this about is incredible, it none the less has a definite high possibility of probability in my book.

Now where am I gonna find a semi trailer full of FPGAs, and where am I gonna hide it from my wife?

Robots

Robots! that is right people the word of the day is robots.

I was trying to come up with an idea for a robot body the other day, when I had an revelation. Ever one always goes for the really cool drone type, or bug type robot, then they attach a remote control and wham we have another expensive toy.

So my idea was to spend as little time as possible on the body, and really make sure my robot was going to be autonomous. I've been dabbling around with an OpenCL+fann library and also with ROS, they are available through a Google search if you really are interested. So after toiling  for a few hours and sacrificing a few toys, I give you Nubby the touchy robot.

This will serve as a test bed for sensory organelle as well as being a test bed for me and neural networks in general.

heh heh heh...robots

Rockbox Robot

I've been day dreaming off and on lately of starting a project to design robots with mp3 players for brains. Most players that run Rockbox have at least a com/usb port and of course audio capture/audio output.

The Com or USB port would be for sending and receiving  information to and from the "body" of the robot. The disk capacity can store instructions and collected data. These two things are simple but, the real innovation would come by using the audio input/output system of the mp3 player, along with some home brew transceiver circuits and some guru level code, as a wireless communications channel. And perhaps I could strap it all on top of a Hex bug. Honey, you now know what I want for my birthday.

But for now just a dream...