Could the AI revolution jump-start economic growth, yet make you poorer?

Open this photo in gallery:

What was Canada’s most common job at the time of Confederation? Farmer.

The 1861 census says 343,000 people, or 41 per cent of the work force, were farmers. And that was probably an undercount, missing the unpaid work of wives and children, or any part-time farm work from the country’s 211,000 “labourers including lumbermen.”

Fast forward a century and a half. Canada’s economy has more than 20 million jobs, but just 257,000 are in agriculture. Farming has gone from the main occupation to barely 1 per cent of the labour force, and falling. Yet the country’s wheat harvest in 2022 was more than 40 times larger than the harvest of 1860.

What did that? Technology.

Which brings us to the hopes and fears for what is expected to be the next big transformative technology: artificial intelligence.

AI could do for the economy what the internal combustion engine, machinery, pesticides, irrigation and new growing techniques once did for agriculture – sparking higher productivity, causing gross domestic product to grow faster and leaving everyone better off. That’s the optimistic story about AI.

Or we could end up in a less rosy future, where AI still sparks higher productivity and greater wealth, yet most of us end up worse off.

Let’s start with the optimistic scenario.

A recent study from Goldman Sachs finds that AI could trigger a 1.5-per-cent annual increase in productivity over a 10-year period, leading to a US$7-trillion jump in global GDP.

It’s not implausible. Since the Industrial Revolution, we’ve been inventing more and more ways to replace human (and animal) labour. A single farmer with a combine harvester can do the work of dozens of 19th-century workers. That’s how an economy can grow faster than the population – something that didn’t really happen before the Industrial Revolution – and how living standards rise.

Two centuries of gains from machines and computers eliminated millions of jobs – don’t advise your child to pursue a career in, say, digging ditches by hand. But it also created even more new jobs and types of work, including new professions.

A recent study of the U.S. labour market published by the National Bureau of Economic Research finds that most new jobs over the last 80 years have come from new classifications – everything from “engineers of computer applications,” a title first measured by the U.S. census in 1970, to “conference planner,” introduced in 1990. Most managers once had a secretary, who took dictation and typed up letters. Then computers came along. The number of jobs in the economy kept right on rising.

That’s the history of creative destruction and economic growth through new, labour-saving technologies. Some kinds of work are eliminated; new kinds of work are created. Output increases and the average person ends up better off.

It’s a positive story, and it’s how the AI revolution might go.

But some more recent economic history offers a different story, pointing to the possibility of a less rosy future.

Since the 1980s, the developed world has experienced a widening gap between economic growth and median incomes. The two had long marched upward in lockstep: GDP growth boomed in the decades after the Second World War, and wages boomed along with it. In fact, pay from middle-income and low-income jobs often rose faster than those at the top end, leading to lower inequality and a narrowing gap between rich and poor.

But a few decades ago, things started to change. There’s been what the Organisation for Economic Co-operation and Development (OECD) describes as a “decoupling of labour productivity growth from real labour income growth.” The result is a “decline in the labour income share” in most wealthy countries, especially the United States.

In plain English, GDP in most rich countries has grown faster than wages. The issue isn’t job loss; it’s job quality – and pay. Jobs are plentiful, but a lot are low wage.

That’s one of the possible results of the AI revolution: inequality on steroids.

Imagine AI delivering a hefty boost to productivity and GDP growth, just as Goldman Sachs hopes, with more wealth being created – but with less and less of it going to workers. More of it would instead go to capital – the companies and shareholders creating and using AI – along with highly-educated workers and managers.

My bet is on this less-than-ideal scenario: higher economic growth paired with higher inequality.

Fortunately, we know what to do about that. Don’t try to stop the benefits of creative destruction; do strengthen the social safety net so that everyone shares in the wealth and nobody is left behind. Make economic growth a win-win rather than a zero-sum game.

Source link