Recently, there’s been a lot of discussion online about AI taking away jobs, but I wonder if that’s really true. A partner at A16z wrote a lengthy rebuttal, offering some quite interesting perspectives.

In short, the theory that AI will end employment is based on the assumption that “the total amount of necessary work in the world is fixed.” But looking at history, that’s clearly wrong. Human needs are constantly increasing, and new industries are emerging one after another.

For example, in early 20th century America, one-third of the workforce was engaged in agriculture. But with mechanization like tractors, that dropped to about 2% by 2017. Despite expected mass unemployment, in reality, labor flowed into new industries, factories, shops, hospitals, research labs, and even service and software industries. Agricultural output tripled, supporting population growth.

The same applies to electrification. In early 20th century America, only about 5% of factories used electricity, but by 1930, around 80% were powered by electricity. Over the following decades, labor productivity doubled. Yet, unemployment did not increase—in fact, jobs in manufacturing, sales, and finance exploded. Labor-saving devices like washing machines and automobiles also created new high-value jobs.

From these examples, we can say that true technological innovation reconstructs the economy and expands the scope of useful work. Did VisiCalc or Excel end bookkeeping jobs? No, instead, a whole new industry—financial analysis (FP&A)—was born, reducing the number of bookkeepers by 1 million but increasing financial analysts by 1.5 million.

What do current data show? According to multiple academic studies, AI adoption has not significantly changed overall employment numbers. However, the types of jobs are shifting. Routine administrative tasks are easily replaced by AI, but analysis, technical, and managerial roles are being enhanced by AI. The demand for software engineers and product managers is rising, and from early 2025, software development jobs are steadily increasing.

Goldman Sachs estimates that the “augmentation” effect of AI is far greater than the “substitution” effect. In earnings calls, the phrase “leveraging AI for augmentation” appears about eight times more often than “replacing existing functions with AI.”

Recent research repeatedly emphasizes that, overall, there is no major change, but a redistribution of jobs is occurring. Researchers at Stanford and elsewhere have found that entry-level jobs are becoming more difficult, while roles where AI provides support are increasing.

On a macro level, there is no statistically significant correlation between AI and unemployment rates. People tend to take jobs enhanced by AI, but in industries with high substitution risk, the trend is reversed. So, overall, it’s neutral but not static.

The emergence of new businesses is also rapidly increasing and strongly correlates with AI applications. The number of new apps on app stores has grown 60% year-over-year. Data sets related to robotics have exploded, jumping from 10th to 1st place in just two years. Knowledge work is still in its infancy, and human ambition remains limitless.

In other words, it’s not simply that AI is taking jobs away, but that the nature and quality of work are changing. Sure, some jobs will disappear, but even more new jobs are being created. Overall, AI is not leading to unemployment but opening the door to a richer era of intelligence.