If an AI takes over the fun parts of your job, do you end up stuck with the boring leftovers?

This explores whether the split of work between people and AI can be redesigned so workers stay motivated, not just productive. The most direct evidence in the collection shows why this is a design problem in the first place. Four experiments with over 3,500 people found that after working alongside generative AI, people felt more in control when they went back to solo tasks, but they were less intrinsically motivated and more bored Does AI collaboration drain motivation when workers return to solo tasks?. The explanation is simple and easy to miss: the AI took over the interesting parts of the task, and the human was left with the dull leftovers. Engagement isn't lost because people resist AI. It's lost because of which parts of the work get handed over.

That matters because of where handing work over is actually happening. Workers mostly delegate AI tasks in information-heavy jobs, and the pattern follows what the technology can do more than which tasks are routine Where have workers actually delegated tasks to AI?. In other words, AI is moving into the drafting, synthesizing and analyzing that knowledge workers often find most rewarding. If the default is "let the AI do whatever it's capable of," the motivation drain is close to built in.

The collection doesn't contain a study that redesigns a workflow and measures engagement, so the honest answer is that this hasn't been tested directly here. It does point to promising tools from a nearby direction, though. Work on agent interfaces describes six ways for humans and agents to interact: planning together, doing parts of a task together, approval checkpoints before the agent acts, checking results, shared memory, and running tasks side by side When should human-agent systems ask for human help?. They were designed to solve a different problem, deciding when an agent should ask for help, because nobody knows the right moment to hand control to a person. Read from a motivation angle, they're also ways to keep people involved in the thinking parts of a task, like setting the plan, instead of only the cleanup. Arguments for keeping AI collaborative rather than fully autonomous point the same way. Humans in the loop catch hallucinations, resolve ambiguity and stay accountable Should AI systems stay collaborative rather than fully autonomous?. Proposals for human-AI research teams make a similar case, with human intuition setting direction while AI explores Can human-AI research teams improve faster than autonomous AI systems?.

What agents still can't do also matters. In a simulated workplace, leading agents finished only about 30% of tasks on their own, and social interaction was one of the main reasons they failed Why do AI agents fail at workplace social interaction?. Single agents also hit limits on work that needs several kinds of expertise and independent checking Do single agents always hit organizational limits?. So the work left to humans could include coordination, judgment and dealing with people, which can be rewarding rather than leftover chores. Whether it turns out that way depends on design choices, not on what the AI happens to be able to do.

The unexpected takeaway is that we may be measuring the wrong thing. Agent evaluation is moving from judging only the final answer to judging the whole process How should we evaluate agent behavior beyond final answers?. Hybrid workflows could take the same step and ask not only whether the output was good, but what the human actually did along the way and whether it held their attention. If teams only track output, they'll keep optimizing in a way that empties out the job.