What Econ 101 Can Teach Us About Artificial Intelligence

Here's why advancing technology often leads to more jobs for humans, not fewer

By Greg Ip of The WSJ. Excerpt:

"The first chart you draw in Economics 101 is the downward-sloping demand curve. It shows that when the price of something drops, people consume more of it.

This elementary rule of economics is remarkably helpful in discussing the effects of technological change. As I wrote last week, when technology makes something cheaper, we consume more of it, often by finding new uses. This lesson repeats through history: When coal-fueled steam power became more efficient in the 1800s, demand for steam power and coal spread. When automated teller machines made it cheaper to operate bank branches in the 1980s, banks opened more branches.

My column cites spreadsheet programs like Lotus 1-2-3, whose arrival in the early 1980s made repetitive recalculation vastly simpler and faster. This reduced demand for bookkeepers but created more demand for people who could run numbers in new and interesting ways, such as accountants and management consultants.

In a recent article for the Harvard Business Review, Ajay Agrawal, Joshua Gans and Avi Goldfarb, economists at the University of Toronto who study artificial intelligence, say that in the 1990s, economists didn’t buy into the hype that the internet and the World Wide Web would upend everything:

“It wasn’t that we didn’t recognize that something changed. It was that we recognized that the old economics lens remained useful for looking at the changes taking place. The economics of the ‘New Economy’ could be described at a high level: Digital technology would cause a reduction in the cost of search and communication. This would lead to more search, more communication, and more activities that go together with search and communication. That’s essentially what happened.”

They apply that lesson to artificial intelligence and more specifically to machine learning, the use of powerful algorithms to make predictions from patterns in large quantities of data. Machine learning, they say, means many problems will be reframed as prediction problems. Autonomous driving no longer involves programming the car to respond in a certain way to a variety of controlled scenarios, but instead, to watch what humans actually do and then respond the same way.

How does this affect jobs? The authors says it depends on whether the technology competes with or complements what you do. Spreadsheets competed with what bookkeepers do (record keeping and calculation) and thus made them less valuable, but complemented what accountants and consultants do (analysis), and made them more valuable.

Mssrs. Agrawal, Gans and Goldfarb say that AI reduces the value of those who predict things for a living, such as whether a mark on an X-ray is a tumor, but it raises the value of those who use predictions to make judgments:

“When prediction is cheap, diagnosis will be more frequent and convenient, and thus we’ll detect many more early-stage, treatable conditions. This will mean more decisions will be made about medical treatment, which means greater demand for the application of ethics, and for emotional support, which are provided by humans.”

So AI will create winners and losers within and among industries and occupations. But what will the net effect be? More jobs or less, and for whom? This isn’t knowable, but history suggests that while a specific occupation or industry may suffer losses, the aggregate effects will be positive. The growth in accounting and analytical jobs since the early 1980s is much larger than the loss of bookkeeping jobs."

See also The Myth of Technological Unemployment: If the nightmare of technological unemployment were true, it would already have happened, repeatedly and massively by Deirdre McCloskey, one of the best economic historians in the world.