AI Impact: Many, Diffuse Winners; Some Concentrated Losers

It’s easy to empathize with recent college graduates concerned about the effect of artificial intelligence on entry-level jobs. At the moment, fewer such jobs (compared to past levels) seem to be a clear AI impact. 

On the other hand, it must also be noted that higher productivity is vitally important for any economy. 

Defined as the ability to produce the same output with fewer inputs; more with the same inputs or more output with fewer inputs, the point is that output, divided by inputs, does matter. 

And that matters because, over long periods, productivity growth is what raises living standards.

In other words, if one cares about higher living standards, one has to care about productivity. 

Productivity Raises Real Incomes

Without productivity growth, higher wages tend to create inflation rather than greater prosperity, as the same output then comes with higher input costs. 

Historically, the countries with the highest sustained productivity growth also achieved:

• higher wages

• shorter work weeks

• better healthcare

• more education

• longer life expectancy

• greater household consumption.

Consumers also benefit because productivity tends to reduce the real cost of goods and services:

• agriculture mechanization reduced food costs

• container shipping reduced transport costs

• semiconductor advances reduced computing costs

• software automation reduced transaction costs.

So luxuries often then become mass-market goods. In other words, productivity converts scarcity into abundance; expands consumer choice and product quality; saves time and reduces transaction friction. 

At the national level, productivity increases economic performance and all the benefits that flow from a robust economy. A society with fewer workers can still grow richer if each worker becomes more productive.

So, ultimately, productivity determines how much people can consume, save, invest, educate, heal, entertain, and invent. 

And the truth is that “productivity,” by definition, means wringing more output from any given level of inputs. And labor is among those inputs. 

So, again, it is easy to empathize with young college graduates who see AI as a threat. But productivity enhancers always represent such threats to someone, somewhere, by direct implication. 

The whole point is “producing more, using less.” 

But it is a somewhat nuanced issue. 

Productivity gains from AI and automation can hurt some suppliers of inputs, especially workers in tasks that are directly substituted, but they do not necessarily imply a net negative outcome for labor or other input suppliers on a societal level. 

Automation can displace some tasks and compress wages in some roles, to be sure.

However, if lower costs reduce prices enough, demand can rise, which can offset some or all of the labor reduction through expansion of output and new jobs. 

In other words, the key question is not whether a task is automated, but whether the market for the final product expands and whether labor is a substitute or a complement.

“Higher productivity” often, or virtually always, does mean benefits at the societal level. 

But some particular suppliers of inputs people) might not benefit in an immediate sense of protected, stable, well-remunerated employment, just as some suppliers of other physical inputs might suffer (buggy whip manufacturers after autos displaced horses for transport). 

Being a “Luddite” is not irrational. It is understandable. But resistance to powerful new automation technology of any type is ineffective in the long run. 

Productivity is overwhelmingly more powerful in the broader sense, in the long run, for the whole of society. 

In other words, there are many diffuse winners, but some concentrated losers.