July 15, 2012

Man Versus Machine

We humans seem to have a knack for creating technologies that threaten our dominion "over all the earth, and over every creeping thing that creepeth upon the earth." The most obvious example would be nuclear weaponry, which spawned the ultimate acronym: MAD, or mutual assured destruction. You could call this the big-bang approach to extinction. A less obvious example would be fossil-fuel burning technologies. This is more akin to sitting in a garage while running your car engine. Either way, the end result is the same: mass extinctions and a great deal of suffering, human or other-wise. It's just a matter of time.

But the more subtle persistent threat to our continued status as top dogs on the planet is the computing machines we have been perfecting. The threat comes in two forms. First, we have become utterly dependent upon them, as the inevitable cyber-attack we all can see coming will demonstrate. Without computers, our monetary system would collapse in hours. Hell, computer-driven stock trading has already made a pretty big dent in our economy.

The second threat is more indirect. Computers can already out-perform humans in an increasingly wider range of activities. And we are busily perfecting self-organizing and self-assembling computers and machines--swarmbots and swarmanoids--that can design and build themselves. The day is not far off when these machines will be linked together, in a paradigm envisioned in the Terminator series, when Skynet became self-aware. This was quickly followed by an apocalyptic war with the machines.

So what's our edge? In the battle between Kirk and Spock, who ultimately prevails? The intuitive captain or the logical second-mate? The answer might just be is in the theories set forth by an 18th Century English clergyman named Thomas Bayes. His contribution was a mathematical formula to assign value to prior information in estimating the probability that a future event will occur.

If you think I'm going to explain the intricacies of Bayesian statistics, think again. I couldn't even if I wanted to. The gist of it is that we humans have evolved a capacity to learn from experience. From infancy onward, we have a pronounced ability to base future choices on prior knowledge.

Till now, computers have approached things on a binary level. Zero-one. Just as DNA builds our bodies from combinations of four nucleotides, so computers build our modern world through two electrical charge states, this and that, zero and one. Spock here.

Our minds are built to operate differently. We make inferences based on probabilities weighted by evidence from experience. Our eyes are a good example of this. We can catch a glance of something and immediately our brain assesses what we saw against what we have already seen and literally fills in the blanks for us. The result may not be exactly right, but most often it is right enough for our purposes. Another example is our capacity to adjust to changing circumstances. A computer can only handle what is has been programmed to handle. We humans can think outside the box, a skill largely dependent on Bayesian reasoning. Kirk here.

But that's changing. Scientists and computer engineers are going all-out to incorporate the principles of Bayesian reasoning into our computers and robots. As the power of micro-processors increases and their size decreases, machines within the capacity for Bayesian reasoning are more and more within the realm of possibility. Eero Simoncelli, a computational neuroscientist at New York University, believes that "when we finally figure out how some of these circuits operate in brains in order to accomplish these feats, we’re going to change engineering. We're going to revolutionize the way we think about designing systems."

Spock and Kirk will finally be joined in one brain. Maybe that will be our salvation, the ultimate man-the-toolmaker escape trick. Or maybe we will have succeeded in creating a new form of life, one that will eventually leave us in the dust. For a peek at what lies ahead, check out the video below:


No comments:

Post a Comment