Here are a couple of conversations I’ve been involved in recently. Does any of it sound familiar?

Conversation with a retired engineer. He asked, “What do you think about this driver-less cars business? Has anybody even considered how many people this will throw out of work? We have to do something about this! Otherwise, joblessness will go so high that it will cause a depression.”

Conversation with a middle-aged family doctor/real-estate investor. He said, with complete confidence, “I’m cutting back on my medical career and moving into real-estate development. Long before I want to retire, I expect my job to disappear due to competition from AI (artificial intelligence).”

When listening to predictions, especially gloomy ones like these, keep in mind that nobody can consistently predict the future. Also remember that the most widely accepted gloomy predictions are especially prone to fail. That’s because people, as individuals, react to and prepare for predictions of doom. They work on the problem before its predicted arrival time. Sometimes they offset it entirely.

Y2K was the ultimate example

The ultimate example came on the first day of this century, with the non-arrival of the so-called “millennial bug,” or Y2K for short.

In the late 1990s, computer consultants warned that at the stroke of midnight on December 31, 1999, computers around the world would freeze up because of a problem with their data-storage limits. Computers used to use just two digits to designate a year. So they wouldn’t be able to tell what came after 1999; ‘00’ could mean 1900 or 2000. The problem had a simple fix, however. By the last day of 1999, most computer owners had attended to it. Damage from the predicted crisis was negligible.

Today’s predictions — gloomy and hopeful — revolve around the expectation that computer speeds will continue to rise, and computer costs to drop, at much the same rate as they have for the past half century.

This trend has led to exponential growth in the processing power of computer chips, coupled with an exponential drop in their cost. This leads to casual-conversation predictions like the two I mention above: artificial intelligence will soon lead to legions of unemployed taxi, truck and bus drivers; and legions of unemployed family doctors will follow soon after.

The logical flaw here is that exploding computer power at shrinking cost is a technological advance. But there are social, legal and practical limits to how quickly business can translate these technological gains into real-world progress (or problems, depending on how you look at it).

Computer makers don’t need government permission to raise the speed of their chips. In contrast, makers of driverless cars face all sorts of problems, long before they make any money.

The shift to driverless vehicles will happen gradually, over a period of decades. After all, driving in traffic involves far more surprises than a champion Go player faces on the playing board. Drivers have to deal with changing weather, full sunlight and deep shadow, unpredictable human drivers with varying skills, unpredictable pedestrian web surfers, potholes, snow-covered street markings and so on.

The shift from human to AI doctors will occur at an even slower pace — in line with how long it takes to earn a driver’s license on the one hand, and a medical license on the other. AI will replace family doctors some time after it replaces the voice and chat help lines that people use when they have a problem with a computer, a cell phone or a utility bill.

Assume technological process leads to economic progress

People have a long record of guessing wrong about the impact of new technology, and on how long it will take for the new technology to become part of daily life. You’ll guess right much more often if you just assume that technological progress eventually leads to economic progress. Continue Reading…