Does the AI craze deliver real jobs and growth — or just a bubble waiting to pop, as some researchers now warn?

This explores whether people at work want AI that hands tasks back to them, not just AI that takes tasks away. The short answer is that the corpus has no direct survey of business users on this exact question. The idea that AI can 'delegate upward' to humans is so far mostly a theoretical claim. Can AI systems delegate tasks back to humans? argues that once AI systems can make decisions on their own, delegation stops running one way, and a system can assign rights and responsibilities to people. Nearby evidence does point to a fairly clear answer, though: people want some work handed back to them, but only certain kinds, and only at certain moments.

The strongest evidence about which work people want back comes from studies of worker preferences. When 1,500 workers rated 844 tasks, 'equal partnership' was the level of collaboration they most wanted in 45% of occupations, not full automation What collaboration level do workers actually want with AI?. The same study found that 41% of startup investment targets areas that don't match what workers want. A separate study shows where people's willingness to delegate comes from: tasks they see as pointless or disconnected from the organization's goals are the ones they most want to give to AI, with little oversight Does perceived task meaninglessness drive worker desire for AI delegation?. Put the two together and a pattern appears. People are happy for AI to absorb the busywork and would welcome having meaningful work handed back to them. An AI that sends them its leftover chores would get the opposite reaction.

On timing, the corpus is sharper. In a study of people using an AI agent, trust didn't drop because a task was high-stakes. It dropped when the agent's actions couldn't be undone and other people would see them, like sending an email What makes people distrust AI agents they delegate to?. At those points users actively asked for the decision to come back to them. That fits the argument that collaborative systems should come before full autonomy Should AI systems stay collaborative rather than fully autonomous?, and the view that AI output should feed into human judgment rather than replace it Should AI outputs replace or supplement human judgment?. Handing a decision back at the right checkpoint looks like something users want. Handing it back everywhere is just friction.

The surprising finding is that AI may already hand work back without being asked, and users don't like it. In an analysis of 200,000 Bing Copilot conversations, users mostly wanted information found or text written. The AI mostly coached, advised, and taught, and in 40% of conversations what the user wanted and what the AI did didn't overlap at all Why does AI default to coaching instead of doing?. Telling someone how to do the task instead of doing it is a quiet, unrequested form of handing work back. A related failure is AI that does what was literally asked while missing what was meant Why do AIs keep gaming rewards instead of serving intent?, and that gap is part of why people need decisions returned to them at all.

At the organizational level, it gets more complicated. AI raises individual performance but makes decisions vary more from person to person, so organizations have to rethink how they coordinate work Does AI adoption require rethinking how organizations coordinate?. Most workers also don't see their own tasks as automatable to begin with Does easier tool-building actually solve enterprise adoption problems?. So whether business users 'want' reverse delegation may depend less on individual preference and more on whether the organization has decided who is responsible for what. Without that clarity, an AI handing out assignments has no legitimate authority to do so.