My goal is to ease fears of a Skynet-like system and, naturally, to discourage the creation of such a system. Skynet was made up of computers running specialized hardware that could learn. That is, it contained circuitry that had not yet been seen. Circuits are made up of logic gates which run on a principle of inputs and outputs. The input would need to be some sort of dataset or even a smaller database from which the circuitry could learn. The output, of course, would be the knowledge gained. Assuming one could find such an input source, we are talking about the output being equal to the natural logarithm of the input. Such a function would be computationally expensive. This is because as the circuitry learned, it would draw power at the same rate. Such a system would crash as it began the boot process.
My goal is to ease fears of a Skynet-like system and, naturally, to discourage the creation of such a system. Skynet was made up of computers running specialized hardware that could learn. That is, it contained circuitry that had not yet been seen. Circuits are made up of logic gates which run on a principle of inputs and outputs. The input would need to be some sort of dataset or even a smaller database from which the circuitry could learn. The output, of course, would be the knowledge gained. Assuming one could find such an input source, we are talking about the output being equal to the natural logarithm of the input. Such a function would be computationally expensive. This is because as the circuitry learned, it would draw power at the same rate. Such a system would crash as it began the boot process.