Artificial Intelligence

Tenets: PROTO, hier

Central concept: If you build a prototype early enough, you may learn from what the machine tells you.

After reading some of the other discussions in this chapter, you may be wondering why we're taking up the subject of artificial intelligence (AI). After all, isn't AI old boring stuff left to tenured professors and geeks too geeky for mainstream geekhood? Haven't all of the good Sci-Fi books about AI already been written? The AI crowd has been telling us for a very long time that computers are going to be smarter than we are someday. Most of this has been laughed off as a techno dream. Someday is still sometime in the future. The AI community has never really delivered anything of value, save for IBM's Deep Blue^[9] and other chess-playing marvels.

I am here to tell you that the AI community will have its day in the sun. It may be sooner than you think. And you may not like it, because when you're dealing with someone or something smarter than you, you may not realize when that someone or something is in control.

First, much of AI theory focuses on the quest for ways to traverse massive decision trees. That's the idea behind the chess-playing programs. The decision tree for the game of chess is a finite, albeit very large, one. Older chess-playing computers could only search decision trees a few plies deep. IBM's Deep Blue used an IBM SP2 32-CPU RS6000 parallel machine with 256 chess accelerator chips to examine 100 million positions per second and perform selective searches up to 30 plies deep.

Remember the idea of next _____'s machine? That's right. Someday we'll have the power of Deep Blue in our wristwatches. I can hardly wait. Then Kasparov will be able to carry Deep Blue everywhere and really figure out why it beat him.

Second, developers have found that in prototyping AI programs early, they occasionally discover that the programs themselves teach them something new. Developing AI software then becomes an iterative process involving not the end user, but the computer itself. Obviously, the more often one does this with progressively faster machines and more intelligent programs, the smarter these programs will become.

Of course, this increase in AI isn't without its perils, as the Sci-Fi authors have reminded us for so long. To see a live example of how this technology can be used for good purposes (at least from a business perspective), visit the Amazon.com site. Notice the number of ways that Amazon tries to persuade you to buy books related to any book you've purchased, any book you've looked at, any book you wished you owned, any book your friends have recommended, any book your mother purchased, any book of yours that your ex-girlfriend burned, any book that your spouse found in your briefcase, any book your dog ate, any book you've read in the bathroom, any book your kid brother borrowed and never returned, and so on.

This kind of AI, one that seeks patterns in human behavior by matching keys between two or more disparate databases, is a relatively new trend. Due to its enormous commercial potential, it will likely gain ground as a hot technology in the coming years as companies seek to maximize their profits. So far, companies like Amazon are exploring the obvious connections between databases. It will get more interesting, however, when AI heuristics begin to point out the not-so-obvious connections. It's fun when Amazon offers you a cool book on Linux-based music composition. It might not be fun when the supermarket has figured out the exact day to raise the price of your favorite coffee.

In this chapter we have looked at several examples of how ideas from the Unix philosophy have infiltrated other technologies. The discussion is by no means comprehensive, and I apologize if I have missed your favorite technology. There are so many fascinating ones out there today that it would have taken yet another book to discuss them. Even then, such a tome would be incomplete, as new technologies come along every day.

My desire is that you will now be able identify those ideas that are Unixlike by nature in any endeavor. As you see the patterns emerge, you will be better equipped to nudge them in the direction of openness and community, rather than seclusion and exclusivity. An idea shared is worth two kept in the mind. We all benefit when one shares one's ideas and we remain open to the ideas of others.

This chapter also serves to illustrate another important point: Unix versions may come and go, but the Unix philosophy will endure for a long time. This is especially true of Linux. Linux will most likely become the version of Unix that fulfills the prediction that Unix will be the world's operating system of choice. Even as Linux overcomes its rivals and enemies through openness and inclusion, Linux may see itself superseded by a newer operating system someday that embodies the Unix philosophy in ways that we couldn't possibly see today. This, too, is okay. When that happens, those who already know how to "think Unix" will be ready for it.

Go forth. Do one thing. Do it well. In the end, what you accomplish will be greater than the sum of all that you do.

^[9]See http://www.research.ibm.com/deepblue.