Originally Posted by Imglidinhere
I don't think you understand the idea of how quickly technology has evolved in the past decade alone. Ten years ago if you told someone that they'd have a casual tablet in the year 2016 that was more powerful than anything that they had in 2005, they'd probably think you're crazy. I mean, my Fire HDX 7" is more powerful than the Pentium 4 Dell I owned back in 2006 and this tablet is only from 2013! Take one from today and you'd see something on par with a Core 2 chip more than likely... if not more powerful yet, and I'm speaking strictly ARM CPUs here, not the ULV Intel stuff that's competing.
A.I. is something that'll likely happen within the next decade or two at MOST, if we don't nuke ourselves into the stone age.
The strides in machine learning we have made in the last five years gets everyone excited. They see the self driving cars and automated factories. That is small fries compared to real AI. Even programs like Watson, even though they are impressive, are leaps and bounds away from being conscience. In the computer crowd it was once thought that we would need a computer capable of human level reasoning and thinking to beat another human at chess. Look how that turned out. We even have one that beats us at jeopardy and it is nothing like a real AI would be. We are decades away from true self aware programs or hardware. To put this into perspective: Lets say we want to run a human brain simulation to achieve an AI. We would need around an exaflop. Right now the worlds fastest computer runs around 33.6 petaflops. We would need around 30 of these to even have the power to simulate a human brain. Then we would need the understanding and the hardware to be able to simulate and scan an active brain into the system. This is only one of a few ways to get to AI, but its one of the easiest. The others would take decades of research and development to even get off the ground. It is nothing now to program a computer to recognize / learn road signs or colors and have a machine respond in certain ways, but getting a program to actually learn and understand the way we do is way more complex. Thousands of times more difficult.
If you're interested I suggest reading Superintelligence Paths, Dangers, Strategies by Nick Bostrom. It is a very interesting read. He goes over almost anything you'd want to know about developing AI. Reading the book has me convinced that we will have AI before the turn of the century, but you might be getting your hopes prematurely up for the two decades mark.