AI was supposed to cut routine work and free up people’s time. Instead, in the tech teams studied, it is raising the amount of work people feel able to take on. Developers using AI are releasing more software, but they are also working longer hours and spending more time fixing mistakes after launch.
When AI cuts the effort needed to start a task, draft a reply or test an idea, the time saved does not always stay free. It often gets absorbed into more tasks, more expectations and a stronger sense that there is still more to do.
UC Berkeley Haas researchers found exactly that pattern in an eight-month study at a 200-person technology company. Employees worked faster and took on a broader range of tasks, often without being asked. Their work stretched across more of the day, spilling into lunch breaks, meetings and evenings, because AI made it easier to keep tasks moving at any moment. Some of the natural stopping points in the workday started to disappear. Over the eight months, what looked like momentum at first gradually became harder to sustain.
That makes the extra work a management issue rather than a personal weakness. When work is always easy to restart, ambitious people are less likely to stop. And once they can do more in the same hour, the pile of pending work stops looking fixed and starts to look unlimited. More becomes possible, so more gets attempted, which is why in many teams AI intensifies work.
The effect reaches beyond software, because it changes how pressure moves through an organisation. If AI speeds up the early stages of a task, many firms will treat the gain as spare capacity and refill it at once with more demands. Speed improves, but relief never arrives. People can produce more and clear more, yet that may only mean the human limit is being pushed further out. In that case, part of the gain from AI is paid for by people working more intensely. Over time, that can raise fatigue and leave strong people feeling constantly behind, even while their output rises.
