The Real Shape of AI's Impact on Work

In a bet that now looks prescient, Google's James Manyika wagered against AI optimists who predicted that within two years, 50 percent of white-collar jobs would vanish. When the deadline arrived, he asked them to settle up. Their response: "No, no, no, we'll reset the timeline." The actual U.S. job numbers tell a different story — jobs are up, not down. Manyika, who has spent years studying how technology changes work, argues that the conventional alarm misses the more interesting and nuanced reality of what AI is actually doing to employment, wages, and the nature of human contribution.

Three Things Happen at Once, Not One

Manyika's central framework is that when AI intersects with work, three effects occur simultaneously — and the proportions among them determine the outcome. First, some jobs will decline because their constituent tasks become easy to automate. He points to roles involving routine tasks like testing code or looking for bugs. But that's only one part of the picture. Second, jobs will be gained in two ways: existing occupations whose demand actually increases (electricians, for example), and entirely new occupations that didn't exist before. Manyika cites research from David Autor at MIT showing that roughly 60 percent of jobs in the U.S. economy today didn't exist in the 1950s. "It's kind of like if you had looked