Tuesday, February 22, 2011

Rise of the Machines




My family and I watched Jeopardy the other night, where the two greatest Jeopardy champions of all time were challenged by Watson, a really, really smart computer made by IBM. You've probably heard by now that Watson kicked butt. Silicon triumphs over carbon.

I was trying to impress upon my kids the historic nature of what they were seeing, as computers rapidly approach the intelligence level of humans. It won't be long until you won't know if you're talking to a computer or a human over the phone.

This point-of-no-return is sometimes referred to as the "singularity." Computers will understand the subtleties of human speech, even humor and sarcasm. They will make jokes themselves. We will interact with them the same way we interact with people, only the computers will have vastly superior memories, and they will keep getting smarter.

What really kicks things into gear is when computers can improve themselves, because, unlike biological improvement (i.e. evolution), which can take eons, iterations of technological improvement can happen very quickly. All bets are off when this happens, but it's a reasonable assumption that robots may become indistinguishable from humans at some point. Debates will flare up over what "rights" should be accorded to these creations. I see the ACLU getting fired up about this.

If all this sounds like the musings of a teenager who has been watching too much Star Trek, I know. It does. But it doesn't make it any less inevitable.

Is this all a good thing or a bad thing? Depends who you ask. Ray Kurzweil is perhaps the scientist/futurist who has given this the most thought. I highly recommend his book, The Singularity Is Near.

Kurzweil is very excited by the singularity. He says it will allow humans to transcend biology, which, as he points out, is already happening anyway. Think cochlear implants, pacemakers, and artificial hips. How much of us has to be artificial before we are...something else?

Computers are also augmenting our intelligence. Ever settle a dinner table argument by looking something up on your smart phone? There you go.

In fact, smart phones are a development worth pondering. I argue that people in the distant future would love to travel back and see today's world, just as we might love to travel back to see when some caveman learned how to make fire or when Gutenberg first fiddled with movable type. Why? Because this decade marks the first time in human history when essentially everyone started taking their computers with them everywhere. Computers have become an extension of ourselves. The other day I ran a short errand and forgot my iPhone. I had to restrain an impulse to run back and get it. A few years from now, it will be unthinkable.

There's no going back. From this point on, computers will be with us wherever we go, and they will become integrated with our daily lives to the point where going off-grid will be the equivalent of losing one's sight or hearing. Computers become a sixth sense, if you will. One that connects us to everything else.

The integration will start to become physical. For instance, instead of having to carry a device around, perhaps it projects on to the back of your retina. So imagine, for a moment, you're at a party and someone approaches. You know you've met him before but you can't place it. Damn!

Not a problem. A micro-camera implant scans the face and immediately recognizes it, having tagged it previously. Then you see an instant summary of who the person is and how you met them. Inside your eye.

I don't know about you, but I'd find that pretty useful. And I'd no longer have to grab my wife's elbow and whisper, who is this guy again?

Kurzweil, the optimist, predicts this will take us to some pretty cosmic places. Frankly, if I just blurted them out here, they'd seem ridiculous. If you let him build his case, though, most of his conclusions seem almost inevitable. He is nothing if not meticulous in constructing his argument.

Kurzweil's principle insight is that numerous technologies are advancing at exponential rates. Moore's Law, except not just for microprocessors. He argues that exponential growth can have some wildly profound implications, but we don't see them because our minds are programmed to project linearly. That, and exponential growth isn't so noticeable at first.

Consider the Chinese legend of the chessboard. Some clever fellow introduces the game of chess to the kingdom and the emperor is impressed. He is asked what the clever man would like as a reward, to which the man says one grain of rice doubled on each square of the chessboard until the board is full. No problem, says the emperor. One piece, two, four, eight...hey, this guy's not so clever, after all!

Probably somewhere around the middle of the third row it dawns on the emperor that maybe it is he who is not clever. By the end of the third row, he owes 16,777,215 grains of rice for a single square. By the end of the last row, he owes 2 to the 65th power, or more rice than there is in the world.

I think he chopped the guy's head off for making him look stupid.

But the point is, exponential growth can be extremely uninteresting for long periods of time. Then, quite suddenly, it becomes very interesting, indeed. According to Kurzweil (and others), we are at that point right now. Fasten your seat belts.

There are others in the uber-nerd community who have a dystopian take on this. Their godfather is probably Bill Joy. They believe that exponential growth is replete with dangers that can't even be imagined. A lone scientist creating a world-killing plague in his bathtub. Smart machines deciding they don't need us anymore. That sort of thing. (I call that last one the Terminator Scenario. I'll be back.)

I guess the main point here is that the nerds don't seem to be disagreeing on the basic point of exponential growth, merely on its implications. Either way, those implications are profound, and they are in our lifetimes, so it is important for anyone with even the least bit of intellectual curiosity to educate themselves on the basics. Listen to the nerds. Entirely new industries will be born in months and others swept away. It will pay to be on the right side of these trades.

I recommend starting with Kurzweil's book and also Andy Kessler's Eat People. And, to quote one of the vanquished Jeopardy contestants, "I, for one, welcome our new computer overlords."

2 comments:

  1. Yes ... But: Human intelligence is deeply linked with human emotion and experience. Winning at Jeopardy is not quite so impressive as writing Hamlet or composing Don Giovanni.

    ReplyDelete
  2. This comment has been removed by a blog administrator.

    ReplyDelete