Artificial intelligence (AI) is automating jobs previously requiring human intelligence.  Some experts predict that AI will soon do most computer jobs.  Should we halt the deployment of AI to prevent a jobs apocalypse?

Let’s first consider the impact of the elimination of professions on individuals, which economists often gloss over.  Our market economy is based on voluntary exchange.  We must trade for things we want because producers will not give them away.

Our economy also involves the division of labor.  We specialize in one thing and trade for everything else.  More precisely, most people specialize in one narrow task, like keeping financial accounts or teaching college economics.  Specialization’s increased productivity allows modern prosperity.

Yet we are in trouble if our narrowly focused skill is no longer needed.  By contrast, farmers can always feed themselves.  If our profession disappears, our standard of living will likely plummet.

Learning a new skill only modestly helps.  That people typically earn the most between ages 45 and 55 indicates the time necessary to become highly productive in a profession.  An accountant or lawyer forced to retrain mid-career may never achieve those high earnings in their new profession.

Although we almost certainly will still need some accountants and lawyers, the disappearance of so many of these jobs so quickly will be huge.

How can such job destruction be beneficial?  Economists respond with scarcity: the resources needed to produce goods and services are limited.  There will always be jobs, and labor-saving technology allows more production.

Because many find this argument unconvincing, let’s discuss jobs differently.  Consider the tasks you want done for you.  Many you will be unable to do, like assembling a car.  Others you could do but would like someone to do to save your time for other tasks.

Because market economy transactions are voluntary, we must pay to have tasks done.  The more time a task takes, the more it will cost.  Tasks requiring significant training will also cost more.  If someone can perform a task in less time, they can be paid less.  The cost of the tools typically needed for faster completion of tasks will offset the labor savings.

Prior to AI, only humans could perform accounting, law, and computer programming tasks.  Mastering the law required significant training (four years of college, three years of law school, years of practice).  Consequently, hiring a lawyer is expensive.

AI will soon (if not already) perform many cognitive tasks for us.  The cost of the AI applications matters for its adoption, highlighting the importance of companies paying the full cost for data centers.

Tasks essentially amount to goods and services.  And AI will automate tasks requiring years of specialized training.   The increased production will raise the standard of living.

People care about their standard of living, not the average.  People not losing their jobs to AI will clearly benefit.  The individuals whose professions are largely eliminated will likely earn less even after retraining.

Several adjustments in markets could offset the impact of lower earnings.  Lower prices from AI productivity will let lower earnings go farther.

The lower cost of AI assisted lawyers will increase the quantity of these services demanded.  Businesses not currently consulting lawyers on certain matters might start doing so.  In the limit, if AI reduces the labor required per unit of service by half and customers want twice as much service, employment may not fall.

Businesses may adjust services in response to AI automation.  ATMs automated many transactions but did not reduce employment as banks shifted people into other customer services.  If people prefer receiving legal or accounting help from a human, full automation will not occur.

Finally, AI may not prove much cheaper than humans.  As mentioned, the apps require data centers and electricity, which could prove expensive.  And salaries could fall in the impacted professions.  Lawyers may accept lower pay to remain cost-competitive with AI.

The human costs from the rapid elimination of good-paying professions will be significant, even if partially offset by these factors.  But these impacts will be temporary.  In fifty years, many fewer people will train to be lawyers or accountants.  After adjustments prevent excess supply, the economy with be left with AI’s higher productivity.

Increased productivity due to AI will yield greater prosperity.  Yet rapid technological change creates significant, if ultimately temporary, costs.  We must make sure the transition costs do not make people reject a technology likely to make our children and grandchildren much better off.

Daniel Sutter is the Charles G. Koch Professor of Economics at Troy University. The opinions expressed in this column are the author’s and do not necessarily reflect the views of Troy University.