Month: April 2016

Our Computer Overlord

AI or artificial intelligence is the concept that a computer can be programmed, or to be more accurate it can be taught, to think, feel, and respond to stimulus as if it were a human. We respond to everything in our lives in a very careful thought out way- even if you don’t realize it. Whether it be the watching for cars as you drive to school, or turning on the heater when you wake up because you feel cold, or even something as simple as feeling the spring push back on your fingers as you type each letter. There is a careful and calculated response your brain has to each of these stimuli. Many are subconscious, but just the same many are taught over years of experience and situations that we have been through.

The thought that an AI can become as developed as a human is simply baffling to me. To be able to program something to even just respond to a conversation cue would be incredibly difficult. Programing the ability to have a conversation is one thing, but to have it be able to pick up on sarcasm, facial cues, emotional tones, and a hundred a one other subtleties that are subconsciously absorbed during an interaction are so incredibly lofty that I do not see a way of doing it. I know there are machine learning algorithms and such out there that allow the machine to ‘learn’ form its experience and be able to respond to them accordingly. But focusing on the conversation example, for it to be able to learn how to converse, we as the programmers must be able to break down a conversation into small blocks so that we can tell the machine how to learn from it. A conversation has millions of millions of ‘moving pieces’ (for lack of a better term) that we would have to break it into and then write a code to understand each part. I think this is just too lofty of a goal to achieve.

However it seems there is a different way of going about this machine learning. What if we could have the machine robot read? All the studies we would have to conduct to understand a conversation would be in scholarly journals and papers. What if we could have the machine read all the documents and understand what they were saying? Could it create its own understanding and therefore teach itself how to respond? Again to me this is beyond my understanding, but I think that is where we as humans are limited. I don’t understand how it could be done, but if I read a bunch of books, and spent years understanding machine learning algorithms and how to program an AI, maybe I could have a rudimentary AI of my own. But if we could develop a computer that would understand basic concepts, could it then teach itself more complex things? Just as a baby starts by learning colors and basic words, could we execute the same sort of logic on a computer, but at a much higher rate because it would not require the down time for development that a baby would. Assuming we didn’t have any memory restrictions, I think this is all possible. Beyond my understanding, yes, but certainly possible.