Tuesday, September 11, 2012

Just a word about the robot uprising

I heard guys on my favorite radio show talking about this today. Here's a video to accompany the discussion, because robots are cool:

So, here's the thing: a lot of people seem to think that because of these advances we've had in artificial intelligence, such as with things like Cleverbot, Watson, iPhone's Siri, and all the videogames that exist, somehow we are closer to reaching a day where our science fiction nightmares come true and intelligent robots start taking over the world.

I'll be perfectly up-front about this: it's not going to happen. Robots can potentially be dangerous if created by dangerous people with bad intentions, but a robot or computer cannot be self-aware and decide to make its mission in existence to wipe out the human race.

A guy who called into the radio show made the stupidest, most herculean leap in logic I've heard from anybody who claims to understand computers. (For the uninitiated, I do have a Bachelor's degree in Computer Science now, so I at least have some excuse for being a boastful know-it-all about this stuff) The caller referenced a computer built back in the 90s which, after being taught the basic rules of Backgammon, was then left to run and play the game over and over until it taught itself how to be good at Backgammon. He then said that, since this was easy enough to achieve, all we needed to do for a robot to become self-aware is to teach it the concept of things, and then it can set about learning all things from the Internet, thus gaining mastery of all human knowledge.

There are two problems with that logic. First, the simpler one: Backgammon is not that complicated. A computer with decent AI can certainly learn to play a board game like that in this way, and it's not surprising. The thing is, the real world has way more rules and countless variables to keep track of. This is why the above robot video is so impressive: the mere act of walking around on four legs is a terribly intricate and difficult process for a computer to handle. It took very talented engineers a long time to carefully design programs that could direct that robot through a real environment. Now, just imagine if the robot had to not only walk around, but use tools, communicate, find paths through more difficult terrain, and even fight, all while remembering that if it turns on its human caretakers, it'll need to find its own source of power to run its engine.

Secondly, the much more damning point: computers, as they exist today, cannot learn concepts. Period. The very notion of teaching a robot the concept of things, or any concept at all, is absurd sci-fi stuff. Everything that we see computers do today that seems intelligent, whether it's Watson schooling people at trivia games, Siri answering our questions, or Cleverbot creepily flirting with us and getting vulgar for no clear reason, all of those things are accomplished through logical algorithms, dealing purely in numbers.

Watson can pull the name Agatha Christie from its database to answer a question, but it has no idea who Agatha Christie is, or even that the letters that make up that name are supposed to signify anything in the first place. It's just following its logic trees and trying to pull likely correct responses to the query; it's just a step up from Googling the question.

At least as they exist today, computers cannot understand a concept, they cannot learn much beyond what they've been programmed to learn, and they cannot have motivations of their own because they have no mind. So, as I said, if some mad scientist wants to create a diabolical killing machine robot army, they may do so, but those robots will never doing anything beyond what they were created to. I won't say that intelligent machines are impossible, but they would need to be created from something drastically different than the simple digital logic machines we use today.

So, that's that. I hope I put a few of your fears to rest. You're a lot more likely to be killed by your dish washer or your toaster than your computer, at least for the foreseeable future. So, relax and enjoy technology, because it's not plotting our downfall. It's here to serve us, just as it was made to.

Peace and love, readers.

1 comment:

  1. Well thank you for that Dan. I feel a lot better about things now. But just suppose.. I mean how would we.. What if....?