Sun - November 16, 2003

It's worth considering these questions now. Robots are already deployed on assembly lines and in operating rooms. Autopilots already operate subways and aircraft. The day when robots can check us in and out of grocery stores, harvest our crops, and drive our cars may be a century away. And engineers and scientists in industrial nations worldwide are committed to creating artificial intelligence sooner than that, if possible.

The poor and uneducated may not feel AI's effects first, but they will be hardest hit by its arrival. In the beginning, robots will cost too much to deploy efficiently. But as production increases and as costs drop, replacing people with machines will begin to look as attractive to companies as moving factories and customer care centers to less expensive overseas markets appears to them today. A vast, displaced class of humanity will be created when machines dominate the world's means of production.

It is tempting to say that such displacement has happened before—during the Industrial Revolution, for example, when thousands of craftsmen lost their jobs to automation. The solution then was to retrain people in new jobs—such as operating or building the machines—or to move them into new roles required by industrial society. Civilization absorbed the Industrial Revolution, just as it has survived and prospered during the Internet Revolution, because new technologies created as many jobs as they made obsolete.

The Robot Revolution will be different. Here's why: Unlike the machines of the industrial revolution which relied upon human operators to guide them and human builders to create them, true AI will be able to function independently. Robots will be able to build robots. Robots will be able to repair robots. Robots may even be able to design robots, but even if they cannot, robot engineering will require advanced training that may not be suited to everyone's abilities and may not be available to the poorest class of society.

The Industrial Revolution created new opportunities in the workplace. The Robot Revolution will be all about supplanting such opportunities. The Industrial Revolution principally affected society's means of production and then spread slowly into other fields. The Robot Revolution—taken to its logical conclusion—means that machines who are as good as (or superior to) human beings will quickly be available to fulfill roles in every field at the same time.

Even if AI with human levels of aptitude does not emerge immediately—and even if some highly skilled, human-oriented professions, such as teaching, policing, acting, psychology, and lawyering take longer to feel the effects of the Robot Revolution—eventually the creation of true AI means that men and machines will be able to perform most functions with comparable skill. Our economy is not prepared for this transition. So things will change quickly. These are a few of the transformations that seem likely:

1. The rich will get richer because their overhead costs of owning and operating a business will fall.

2. For a while, the middle class will prosper as well as the cost of launching and running its own businesses will make business ownership a realistic opportunity. But the market cannot remain saturated forever and many businesses born of the automated labor boom will not prosper.

3. The world will require fewer people to do its work. Government's will therefore encourage and perhaps even mandate birth control to slow population growth.

4. To the extent that jobs disappear and are not replaced, our economy will have to devise new means of distributing wealth or risk creating a dispossessed underclass larger than any the world has ever known.

5. Human beings will confront the mixture of pride and personal insecurity created when parents are supplanted by their children.

Of course, all this may be a bad dream. AI may not come to pass. Or we may prove wiser at deploying it than these shadowy suggestions of one possible future foretell. But because we're a species very much in love with our own self importance and very much convinced of our own adaptability, the pursuit of true AI seems inevitable. The question is: By the time we develop an intelligence equal to our own, will we have first considered whether we want it at all?