Soon the company that hires you will expect your work to amount to nothing.
We currently measure a career by the value of the work it produces. But AI is about to make execution almost free. And when execution is free, anything that clearly works will be immediately copied.
If an agent can build a good product from a prompt, it can copy anything that already exists. Every successful app, campaign, and post will invite thousands of competent clones. The world will fill with things that work—and they'll all look the same.
In a market of clones, execution stops being a differentiator. A familiar idea competes with a thousand alternatives; something genuinely new has the world's attention to itself. Any path that looks promising will be immediately swarmed by agents, so the opportunities left for humans will be the ones that look risky, irrational, or simply wrong. Agents are limited to their training data, but humans create new data just by living: strange frustrations, tastes, relationships, and behaviors that nobody has documented yet. Founders mine those feelings to pursue an idea everyone tells them won't work.
Airbnb thought people would want to sleep in strangers' homes. Dropbox entered a file-sync market investors already considered too crowded. Designers initially dismissed Figma as a toy because professional design software obviously did not belong in a browser. Each opportunity remained available because initially it looked like a bad idea.
Hiring will then start to resemble gambling. Fund a portfolio of 1,000 people pursuing ideas that appear likely to fail, and hope one bad idea turns good. A rare original idea could therefore be worth a billion times more than the thousandth clone of one that already exists.
