Photo by Markus Winkler on Unsplash
 

Artificial intelligence will increase productivity—output per hour worked—in some sectors of the economy, but not others. Every change in productivity triggers other changes in spending and production and employment. AI will lead to fewer jobs in most areas of increased productivity, but more jobs in the low-productivity-growth fields.

Production in two sectors of the economy. Dr. Bill Conerly
 

Simple Example of AI Increasing Productivity

The basic principles can be illustrated with a simple example. Imagine an economy with two products, food and shelter, produced in roughly equal amounts. Their output is illustrated in the first figure nearby.

Production after increased productivity in food. Dr. Bill Conerly

Now suppose that the farmers become more productive, doubling their output, while the shelter sector continues at the old production rate. The second figure shows production, totaling a 50% increase in the aggregate.

The people in this economy can now buy, in total, 50% more products. But most likely they don’t want all of the gain in food. They typically would want more shelter as well as some increase in food quantity or quality. A market economy responds: Food prices decline, increasing the amount of money that people have available for shelter. As they try to purchase more shelter, the price of housing rises. The shelter industry responds to greater demand by hiring more people, many of whom will be those who lost their jobs in the food industry.

Our economy is much more complicated, but it’s still true that employment will decline in most of the sectors with the greatest productivity boosts from AI, while jobs will grow in the sectors that have the least impact from AI. In other words, job growth in the future will favor carpenters more than computer programmers.
 

AI-induced Productivity And Job Losses

Specific sectors that use AI to enhance productivity will generally shed workers but with some exceptions. Two key issues determine the extent of employment change: how much AI drops the price of the final product, and how sensitive demand for that product is to changes in price.

In a few cases, AI will drastically lower the cost of production. Competition among producers will lead to lower prices. The degree of price decline is critical. Right now, the occupations and tasks most likely to be made more productive constitute small portions of total costs. For example, OpenAI’s research suggested that insurance companies would have very high exposure to AI benefits, with half of the jobs made more productive with AI. But their employee costs are a small portion of the total cost of providing insurance. Most of the cost is paying claims. So cutting the insurance company’s labor costs in half would have a relatively small impact on the cost of my homeowner’s insurance. Still, any price reduction will be welcomed by buyers.

In a few fields, such as computer programming, though, the cost decline will be substantial. In those fields, we need to understand how much the price decline will stimulate increased quantity demanded.

Take my business website as an example. My web developer could add many different features, each at some cost. He and I discuss the benefits and the cost of possible features, and I make a decision to buy or not buy. But if the cost were significantly lower, I would go for more features.

Economists call this concept “price elasticity of demand,” the percentage change in the quantity demanded divided by the percentage change in price. Salt, for example, is not very sensitive to changes in price. If AI increased the productivity of workers in salt mines, we’d probably see job losses, because the total quantity of salt production and consumption would not change much.

Sales of some products, though, increase hugely as price goes down. We have seen the price of cell phone data drop substantially in recent years, and as a result, people use a lot more data, streaming video on their phones as well as listening to music.
 

Job Gains From AI

Increased quantity demanded will sometimes drive large employment gains. The Model T was first hand-built. After the assembly line was used by Henry Ford, costs fell. Other car manufacturers used similar techniques to cut costs and lower prices. Ford brought the price of the Model T down from $850 to $260. At the lower price point, millions of people wanted cars, and total employment in the industry soared. Thus, it’s possible for labor-saving innovation to create new jobs. In most cases, industries that employ AI extensively will see reduced employment, but there will be some exceptions.

AI-enabled productivity gains will trigger increased spending elsewhere in the economy, such as shelter in the simple example above. In the real world, look for long-term gains in discretionary spending categories. Food calories probably won’t increase much in the developed world, but people will dine out more, buy prepared foods at the supermarket and use food delivery services more.

They will likely upgrade their wardrobes, limited by precious closet space. They will add recreational equipment ranging from campers to kayaks. They will hit gyms and spas more frequently. They will also spend on other experiences, including concerts, sporting events, and travel.

Consumers will spend more on housing, by up-sizing their homes, upgrading furnishings, and living without roommates.

The discretionary sectors that also lower prices due to AI-enabled productivity gains will enjoy double benefits: greater demand from great household purchasing power, plus lower prices relative to other sectors.

Overall, the greatest change will come to industries with a high concentration of tasks that will be aided by AI. All will see unit sales volume increase, but only a few of these industries will enjoy increased employment. General improvement will spread through most of the rest of the economy.

Mass unemployment due to AI is very unlikely, but these shifts in activity will hurt some people with highly specialized skills. Someone who can work in any industry—such as a janitor or salesperson—will probably make the adjustment easily. Someone tied to a particular part of the economy will be at risk if that sector downsizes employment because of AI-induced productivity gains.

More By This Author:

Recession Forecast Still Correct For Late 2023 Or Early 2024
China’s Economy Looks Weak, Bad News For U.S. Exports
Which Companies And Industries Will/Won't Embrace AI - And Why