I’ve always been curious about this. Even though there are more and more chances to hear the claim that AI will end employment, when you look at the actual data, that view is far less well-grounded than anyone would imagine.

Recently, David George, a general partner at A16z, has offered an interesting counterargument to this debate. His basic claim is that the “end of employment” thesis brought about by AI rests on a fundamental fallacy: the total amount of work is fixed. In other words, the amount of work that should be done across the economy is always changing, and even if new technology replaces old jobs, entirely new areas inevitably emerge.

Looking at history, the validity of this perspective becomes clear. In the early 20th century, one-third of the American labor force was engaged in agriculture. Thanks to mechanization of agriculture, such as tractors, that share fell to about 2% by 2017. At first glance, it may seem that automation led to massive unemployment. But the reality is different. Agricultural output nearly tripled, supporting population growth. The unemployed agricultural workers flowed into industries such as factories, stores, offices, hospitals, research institutes, and—eventually—the software industry, which did not even exist at the time.

The process of electrification was similar. Think about the fact that electrification was not merely a switch in energy sources; it brought factory restructuring and the birth of entirely new consumer goods. In the early 20th century, only about 5% of American factories used electricity, but by 1930, about 80% of manufacturing was powered by electricity. Over the following decades, labor productivity doubled, and new job categories—such as those in manufacturing, sales roles, and credit provision—grew rapidly.

What about the advent of Excel and VisiCalc? Bookkeeping jobs certainly declined. But in the process, an entirely new industry—financial planning and analysis (FP&A)—emerged. So, for a reduction of about 1,000,000 bookkeepers, there was an increase of about 1,500,000 financial analysts.

So, is something similar happening with AI now? The latest academic research offers an intriguing answer.

According to research from the National Bureau of Economic Research (NBER), AI adoption has not brought about major changes in the total number of jobs, but it has begun to reshape the distribution of occupations. While routine office tasks are easier to replace, analytical, technical, and managerial work is instead being strengthened. A survey by the Federal Reserve Bank of Atlanta estimates that, on average, more than 90% of companies did not see any impact from AI over the past 3 years. Research by the U.S. Census Bureau finds that only about 5% of companies using AI reported effects on their employee numbers, with increases and decreases occurring in roughly equal amounts.

Yale University’s Budget Research Office points out that “the impact of AI on the labor market reflects more stability rather than major economic upheaval.”

That’s why today’s data does not support the doomsday view. Instead, new data points in the opposite direction. One of the clearest examples is the rapid increase in hiring software engineers and product managers. In Goldman Sachs’ analysis, executives using the phrase “AI augmentation” did so about 8 times as often as they used “AI replacement.”

Why is that? Because when AI becomes a tool that enhances cognitive capabilities, people start seeking more advanced work. The creation of new businesses is increasing explosively, and the share of new apps appearing in app stores has increased by 60% year over year. For robotics-related datasets, they jumped from 10th place to 1st place in just 2 years. All of this is evidence that AI is not simply taking jobs—it is creating entirely new industries and occupations.

Of course, some jobs will shrink. The U.S. Bureau of Labor Statistics predicts that jobs for customer service representatives and medical administrative staff will decline. However, at a macro level, no statistically significant relationship has been found between AI and unemployment rates or job growth rates. Instead, industries with AI enhancement tend to show higher job growth and lower unemployment rates.

At its core, this is not a new story. Most jobs created since 1940 were occupations that did not exist in 1940. I think the same is happening again. AI may indeed eliminate some jobs, but believing that that’s the end misunderstands human nature.

Humans are always seeking expansion. When food prices fall, spending increases on housing, healthcare, education, travel, and entertainment. The labor market is no different. Human ambition never runs out, so new jobs are constantly created, new frontiers appear, and there is a need to conquer them.

Knowledge work is not the end—it’s the beginning. Automation removes repetitive tasks, and human work is elevated to higher levels. When AI becomes cheaper, human needs and ideas don’t disappear. Quite the opposite. Cheaper AI, a bigger market, new companies, new industries, and more advanced human roles are ahead.

There are no fixed values for the amount of work or for cognitive ability. That has always been true, and it will remain true. AI is not the end of employment—it’s the start of a richer era of intelligence.