Hal Masover | Apr 22 2026 15:00
The 2016 presidential election featured the major candidates competing on who had a more compelling job creation program. That drove me a little crazy. It’s been clear for a long time that the biggest long-term challenge to the United States is a long time labor shortage.
From 1987 through 2007 the US economy, as measured by Gross Domestic Product (GDP), grew by an average of 3.1% per year.
After Barack Obama was inaugurated in 2009, he ordered the Commerce Department to produce annual 10-year forecasts of the US capacity for growth. This was a big deal both because we were at that moment in the middle of the worst recession since the Great Depression, and because prior to then the Commerce Department had only produced 5-year forecasts, so this was a potentially important change that allowed for longer term planning.
But the first forecast released early that year was a rather large disappointment. The Commerce Department forecasted 1.5% growth over the next decade. The #1 constraint to economic growth cited by the report was labor shortages.
So, seven years later when Clinton and Trump were coming out with competing proposals for jobs initiatives, I was practically screaming at the TV, “No! It’s not jobs we need. It’s people!”
And, during the next 15 years, the US labor force grew by only 0.6% per year.
But economic growth has two major components. Population growth and productivity.
It turns out that our GDP grew by a faster than expected 2.2% annual rate. That’s because productivity grew at an average of 1.5% per year.
Productivity and AI: The Main Point
This brings me to the most important point. Productivity.
During the first Trump Administration, their anti-immigration policies made it clear that we were not going to solve our labor shortages by importing people. That left only one potential solution. Automation.
We are only at the beginning of the AI revolution. And already it’s making big changes. I made extensive use of it while writing this article. No, this was not written by an AI. I wrote all of it, but it’s important to get the facts straight when writing and article like this. AI is an amazing research tool. I rarely use Google Search for this kind of work now because Google Gemini can find what I’m looking for in far less time. What might have taken as long as 30 minutes to an hour to dig up, in the extreme, now takes under a minute! That’s amazing!
The impact has the potential to be as dramatic or even greater than the impact made in the 1990s by the twin events of the advent of the internet and putting a computer on everybody’s desk.
We are only at the beginning of the AI revolution and it’s happening super-fast. Chat GPT was first introduced in November of 2022, just a little over 3 years ago.
Chat GPT is a Large Language Model, or LLM for short. It’s not intelligent but the various companies putting out LLMs for public use have given them an interface that makes them seem intelligent.
In the 3+ years since ChatGPT, LLMs have moved from being a curiosity to now fast being everywhere. Personally, I am now using an LLM to search for things as much or more than I used to use the search bar on my browser. But this is just scratching the surface of what these things can do. Last week, I listened as a technology reporter, during a podcast, constructed an entirely new website with basic pages like a bio and even an embedded video game in roughly 2 minutes using Anthropic Code.
My new fitness watch has a built in AI agent that answers my fitness questions partly by using personal data about me from sensors on the watch. It can then generate carefully tailored plans to help me reach my fitness goals.
This goes on and on. For free, I can use any number of AI agents and more and more devices have AI built in to them. And it will only get better.
So, humans have done and are doing what we have always done. Adapt. Faced with chronic labor shortages we are finding ways to continue to advance and grow. We are using LLMs and actual robots to make human workers far more productive than ever before.
As will all such productivity revolutions, many workers will be replaced by machines and most of those will never find jobs as good as they previously had. Humanity will likely be just fine, but many individuals will not.
In 1910, one in ten jobs were on farms. And really, that was a dramatic improvement over earlier civilizations where almost everyone worked at providing sufficient food. If most of a civilization’s efforts are providing enough food, there’s little time or resources to do much else. As the number of workers needed for food production shrinks, that opens up more possibilities for humanity. The industrial revolution was only possible because humanity had reached a point where there were fewer people were needed for food production leaving enough available workers for the new factories. So, the fact that 9 in 10 Americans worked at something else besides food production in 1910 was a big deal. It opened the door for the next revolution, transportation. By the late 20th century, 1 in 7 Americans worked in the automobile industry – an industry that did not exist in 1910. Meanwhile, farm mechanization meant that even fewer workers were needed for food production, and factory automation meant that the stage was set for the next revolution – the information age.
And just like all those other revolutions, the stage is now set for new revolutions. I say revolutions, plural, because AI will open the door for industries we have not yet imagined.
I listened to an interview with Dario Amodei, the CEO of Anthropic. He certainly knows more than almost anyone else about not only what we are doing now in AI, but about what’s coming. He thinks that in as little as two years many of today’s jobs will be obsolete. He may be right. And if it happens that fast, the speed might cause very difficult societal upheaval.
But I do not envision a dystopia similar to The Terminator movies or the Matrix where the machines takeover and humans are no longer needed. Instead, I foresee another period of rapid advancement where people are freed from their current jobs to go on to new higher levels of work.
No one mourns the loss of manual farm labor jobs. No one mourns the loss of typing pool jobs. No one mourns the loss of keypunching jobs. We lost all those jobs and today have a very tight labor market. We will go through a similar evolution. The only question is, can we handle the speed. The transportation revolution took roughly 50 years to unfold. The information revolution, closer to half that. The AI revolution might be even less than half of the time the information revolution took.
An interview I have cited before is worth repeating here. Sometime in or about 1992 I saw an interview with Sir John Templeton, founder of the Templeton Funds and the man who coined the term Emerging Markets. At the time we were dealing with the Savings & Loan crises, and I think we were in recession. Everyone had a list of worries, as we always do.
But when asked about the then current list of worries, Templeton was optimistic. “There’s always something to worry about,” he said, “but life is better today than it was 30 years ago, and I won’t be here to see it, but it will be better in 30 years than it is today.” He was 88 at the time. He died at 100, so he didn’t live to 2022 to see if his forecast was correct, but most people would agree that our standards of living in 2022 were better than in 1992. And it only seems like the rate of improvement in our standards of living is increasing.
The machines are not here to takeover, unless we want them to. And we do want them to take over some things. They are not conscious. We are not living through The Terminator or the Matrix. We have invented yet another labor-saving device in a long series of such inventions going back to the wheel.
Hold on tight, it might be bumpy, but I fully expect the destination to be nicer than where we are today.