To quantify the amount of data a human can process, we'd need to first digitize a state of organic memory and "identify" the processes that it is conducting at that point in time.

By "identify" I mean to break down the problems being worked out into programmable steps which a machine can follow. This should enable us to roughly gauge how much processing power a human brain requires.

The thing about machines is they don't inherently think. They follow simple instructions at high speed to produce a result. In essence, they can be made in such a way that they are self-adaptive and capable of learning. However this is not organic learning. The machine is basically recording pattern as input and compiling a program that they will call up and follow the next time the pattern is recognized. If the pattern is triggered and they respond in a way that is incorrect, they will be given new input to better define the behavior.

The problem with this method is, there is a significant disconnect between the learning process of a machine and that of a human. While both learn through example, a human develops an awareness of situational behavior which is further applied to the basic reaction. A robot can only attempt to recognize the most simple pattern in a serious of examples. Because situations are nearly always varied, the machine must learn to seek out and identify a list of situational indicators in order to determine a proper course of action. Because of the complexity of such indicators and the fact that vantage point is not static, this type of process requires a great deal of work to execute.

Humans seem to have an inverse ability to process compared to machines. Things like math generally take humans a few seconds while a machine can achieve a nearly instantaneous result. However humans can absorb and evaluate a large volume of information in an effective instant, while robots must individually identify every piece of input before it can assemble an internalized data model which can be more easily processed by the machine.

Pic represents a partially constructed data model as a machine might identify flat image input.

>>381
>The problem with this method is, there is a significant disconnect between the learning process of a machine and that of a human. While both learn through example, a human develops an awareness of situational behavior which is further applied to the basic reaction. A robot can only attempt to recognize the most simple pattern in a serious of examples. Because situations are nearly always varied, the machine must learn to seek out and identify a list of situational indicators in order to determine a proper course of action. Because of the complexity of such indicators and the fact that vantage point is not static, this type of process requires a great deal of work to execute.

So a good place to start would be teaching the program practical forms of intent, risk, profit, and probability,etc... right? [pic related, feeling OCD] Given an array of sensory equipment, it can than begin to teach itself, even if it is just making everything a math problem.

I would argue that that's how we do things from the onset, and our only difference is emotion triggered by chemical information pathways instead of purely electrical.

currently a biological brain seems to be as good as it's going to get in terms of storing and processing information. Computers are incapable of doing and being anything more than tools and extensions of human mental processes.