I've always wondered, is it really true that AI will take away jobs? Recently, a16z partner published a lengthy article, and the answer is surprisingly simple.

The root of AI panic is the false assumption that "the total amount of work is fixed." But looking at history, that hasn't been the case. When agriculture was mechanized, American agricultural workers decreased from one-third of the labor force to 2%. Did unemployment increase? No, quite the opposite. Agricultural production tripled, and workers moved to factories, offices, hospitals, and the software industry. Entirely new industries emerged.

The same pattern occurred with electrification. In the early 20th century, only 5% of factories used electricity, but by 1930, 80% were electrified. Labor productivity doubled, but employment didn't decrease—in fact, it increased. Jobs that were previously impossible, like washing machines and automobiles, were created one after another.

Then came the era of VisiCalc and Excel. Bookkeeping jobs were supposed to disappear. But what actually happened? The number of bookkeepers decreased by 1 million, but financial analysts increased by 1.5 million. Salaries also rose.

The same is happening with AI. Goldman Sachs estimates that AI will "enhance" more jobs than it "replaces." At earnings calls, the phrase "AI enhancement" appears eight times more often than "AI replacement."

The demand for software engineers is a prime example. Because AI has made coding more efficient, the number of engineering jobs has actually increased. The number of product manager openings has also reached record highs. When human productivity improves, demand for both roles grows.

Academic research is also interesting. An NBER paper concludes that "AI has not significantly impacted overall employment." The Atlanta Federal Reserve estimates that "more than 90% of companies reported no impact from AI." The Census Bureau reports that only 5% of companies using AI have affected their employee numbers, with roughly equal numbers reporting increases and decreases.

Yale University’s latest data also states, "The impact of AI on the labor market remains largely speculative. The data shows stability."

This is the key point. Overall, the effect is neutral, but not static. Some jobs disappear, some emerge, some decrease in value, and others increase. At the current pace, the demand for developers will surpass pre-pandemic levels within two years.

The example of travel agencies is also instructive. Since their peak in 2000, employment has halved. But overall employment rates remained stable, and the salaries of remaining travel agents increased faster than in other industries. In 2000, salaries were at 87% of the industry average; by 2025, they reached 99%.

In short, while AI will certainly eliminate some jobs, thinking that everything will end because of that is a mistake. The real story is about labor market reorganization, which has happened before. Most of the jobs created since 1940 didn't exist in 1940.

Human ambition is endless. When food becomes cheaper, spending shifts to housing, healthcare, education, travel, and entertainment. The same will happen in the labor market. The emergence of new businesses is exploding—app store new listings are up 60% year over year. Robotics is also a new frontier. Data sets related to robotics jumped from 10th to 1st place in just two years.

Therefore, the future isn't an era of unemployment. Cheaper AI, larger markets, new companies, new industries, and more advanced human work await. AI can be seen not as the end of jobs, but as the dawn of a richer age of intelligence.

Certainly, some occupations will decline. Customer service reps and medical administrative staff are predicted to decrease, and it may already be happening. But overall, the benefits of productivity gains will take time to spread across the economy, and ultimately, employment will increase. Promoting retraining is important, but the data shows that AI isn't destroying jobs—it's reorganizing them. That has always been the case, and it will continue to be.