Shrinking government is one of the hardest tasks on planet earth. That’s why it rarely happens. But Lewis Andrews argues that artificial intelligence (AI) might just do the trick.

Recent advances in artificial intelligence have led many observers to worry that computers will soon replace far more jobs than imagined just a few years ago. The World Economic Forum now predicts that over 85 million positions could be lost to automation by the year 2025, many in law, medicine, accounting and other fields once thought immune to electronic substitution.

Industry experts like IBM CEO Arvind Krishna argue that the worries about this dramatic change are vastly overblown. Like every past technological innovation, he says, AI will eventually create many more employment opportunities than it eliminates, producing jobs in which a person’s productivity will be enhanced by his or her ability to use smart and dexterous machines.

But when it comes to the public sector — those who work for America’s towns, states and the executive branch in Washington — a growing payroll might not, in fact, be the best outcome. Indeed, a good case can be made that, when it comes to government, the best use of artificial intelligence is to replace as many humans as possible.

Can we get an “amen”? Of course, as Andrews immediately adds, the work many public employees do is important and many of them do it well. Those folks aren’t the problem. More than a few of the 22 million people working for governments at various levels, however, are part of the problem — and they’re often unionized and work toward electing Democrats who create more government jobs. It’s an awful cycle.

So, he asks, “What can be done to restore both fiscal and ideological balance to our democratic way of life?”

Fortunately, advances in AI suggest an intriguing alternative: deploy technologies which have been shown to provide quality public services with far fewer personnel, especially in areas like public education which by itself accounts for a full third (7,062,560) of all government workers. If you cannot do away with public employee unions, in other words, at least make them smaller and, as a result, far less influential.

Again, can we get an “amen”? Andrews proceeds to outline other ways that AI is being developed and used, which could provide somewhat of a template for government services.

According to a June 2021 study published by George Mason University’s School of Policy and Management, AI will also make it possible to greatly reduce the number of government workers who perform routine bureaucratic tasks. It notes that the federal government’s General Services Administration has already developed what it calls a “leasing robot,” which manages public buildings in eleven regions across the US and processes more than $6 billion in payments each year. A related paper by Deloitte Government Trends estimates that automation could save the public sector as many as 1.3 billion human administrative hours every year.

Resolving fines and other low-level legal disputes, giving tax preparation advice, performing and interpreting medical tests, making welfare payment decisions, generating instructional publications, supervising compliance with police penalties, conducting background checks, negotiating contracts and processing claims, delivering mail, building and maintaining roads, detecting grant fraud — these are just a few of the many government functions where AI can clearly replace humans. According to Oxford academics Carl Frey and Michael Osborne, around a quarter of all modern public sector jobs could easily be automated by 2030.

When do we begin? Well, carefully, Andrews says.

It is important to understand that using AI to reduce the number of public employees is not about insensitively firing existing workers. With more than 75 million baby boomers entering their retirement years, there is no reason why a thoughtful transition to “smart government” would have to jeopardize any current staffer’s job security.

Nor would it require the average citizen to tolerate an inferior or dehumanizing experience while interacting with public institutions.

He concludes:

At a time when American democracy at all levels is more in the dysfunctional grip of public unions than ever before, our first response to any new challenge should be, “How can technology help us solve this problem without swelling the ranks of government workers?”

Read the whole thing here.