Terminator: Salvation is now in theaters so robots are on everyone’s minds (even if, by the majority of accounts, the movie sucks really bad). Coincidentally, I was out for drinks with some friends last night and got into a heated discussion with one particularly close-minded individual over the future of robots and artificial intelligence.
The argument boiled down, as it usually does, to what it means to be human, and whether machines can be programmed with basic human emotions like desire. My debate opponent said no. Having formed my opinion from talking to dozens of robotics and artificial intelligence researchers and social scientists over the past year, I said yes. I recently blogged about the U.S. government’s EATR project – a robot that consumes other materials for sustenance when traditional fuels are not available. Such a robot will obviously know when its fuel levels are running low, so it will look for a snack (hence the government’s silly name for it). Translation: I’m hungry, need to eat. If that isn’t a rudimentary example of “desire,” I don’t know what is.
It’s important to note that while the above example is neat, it’s still very basic. The point, however, is that that sort of autonomous capability is going to ramp up very quickly. Consider that in 2004, the U.S. government held a road race for robot cars that no one actually finished (the furthest anyone got was 12 kilometers). A year later, five cars completed the race. Car makers expect fully self-driving cars to be on the roads within the decade, and indeed, some current models are already using some of the technology involved, like crash detection.
This is all because of Moore’s Law (the theory that processing power of computer processors doubles every 18 months or so), which actually applies to far more than just computers. As each individual technology improves, the pace of technological acceleration quickens. So, for example, while separate advances in computers, lasers and video projection net some benefits to those individual fields, if you put them together you suddenly come up with new possibilities, like holograms. Robots, in that vein, are nothing if not a collection of various technologies. They are totally integrated systems that benefit from just about every technological advance that comes along.
Last night’s argument really broke down when we started talking about the robot pizza-making machines I blogged about the other day. If you watch the video attached to that post, there are some traditional pizza makers in there saying that a robot will never make a pie as good as a human, a sentiment my debate opponent agreed with. There’s actually nothing more patently false – if the perfect pizza depends on the perfect ingredients assembled and cooked in a perfect fashion, then a robot is far more likelier to get it right every time than an error-prone human is. Similar nay-sayers years ago claimed that a computer would never beat a human at chess, or that robots couldn’t make cars as well as humans, or that computers couldn’t fly and land a plane as well – interestingly, commercial pilots today get reprimanded if they try to manually land their aircraft.
For my part, I’m particularly proud of the perogies I make. Problem is, they tend to turn out differently each time I make them, as in sometimes they’re good and sometimes they’re great (we Polish are not big on writing down recipes). I wish I had a robot that I could program with my method, then tweak it until it got to that perfect method. Then I would have great perogies all the time.