AI assistants often take over the fun, engaging parts of a job and leave humans with the boring leftovers — why does that keep happening?

This explores which parts of professional work people find rewarding in themselves, rather than only for the paycheck, and how AI changes that. The corpus has no study that maps intrinsic motivation task by task. Its clearest evidence comes from the other direction: what happens to motivation once AI removes part of a job. Four experiments with over 3,500 people found that after working alongside generative AI, workers felt more in control when they went back to working alone, but they also felt less intrinsically motivated and more bored Does AI collaboration drain motivation when workers return to solo tasks?. The researchers' explanation is the interesting part. The AI had absorbed the engaging parts of the task and left people with the mundane remainder. So one answer to the question is: the parts AI is best at taking over are often the parts people enjoyed.

To see which parts those might be, it helps to split a job into layers. Narayanan and Kapoor divide knowledge work into deciding what to do, executing it, and delivering it. They argue AI mainly compresses the middle execution layer, while deciding and delivering hold steady or grow Does AI really compress all layers of knowledge work equally?. Put that beside the motivation study and a plausible picture emerges, though it is an inference and neither paper states it. For many workers, the absorbing part of a job is the hands-on making: drafting, building, solving. That is exactly the layer being automated. Judgment and handoff remain, and they may feel more like supervision than craft.

A second candidate is the work of getting better at something. AI productivity gains show up when people apply skills they already have. When workers use AI to learn new skills, those gains disappear and the learning itself suffers When does AI actually boost worker productivity?. If developing mastery is part of what makes work satisfying, AI can quietly cut that loop short. A third candidate is human connection. Romero argues the only jobs that are truly safe from AI are those where people pay for a particular person rather than a deliverable Which jobs will actually survive automation by AI?. That suggests the relational side of work is both what lasts and, for many people, what makes the work feel meaningful.

The evidence does not all point one way. Anthropic's own data shows that people who delegate the most work to Claude are the most optimistic about their careers and believe their skills are gaining value Does delegating work to AI actually damage worker skills?. That data is correlational and drawn from Claude's own users. Feelings about AI also follow a U-shape: workers who were slowed down by AI and those who were sped up the most both worry most about losing their jobs Does AI productivity gain always ease job displacement fears?. Satisfaction with a job and security in a job are moving separately here. You can feel more effective and less engaged at the same time.

The takeaway you may not have expected: deciding what to hand to AI is also deciding what kind of job you keep. Productivity studies rarely measure what is left over for the person. The one study in this collection that did measure it found the leftover work was more boring.