If AI touches every part of a job a little, versus a few tasks a lot, does your company hire less?

This explores whether it matters for a firm's hiring if AI's reach is lumped into a few tasks or spread thinly across everything workers do. The corpus's most direct answer comes from task-level data on firms from 2010 to 2023. Higher average AI exposure reduces a firm's demand for labor. But when the exposure is concentrated, meaning it hits only a few tasks, workers can move to the tasks AI doesn't touch. The net employment effect is then modest Does concentrated AI exposure enable workers to adapt and reallocate?. How much AI touches a firm's work matters less than how it is distributed. The corpus measures concentration across tasks. It does not measure it across individual workers.

The intuition is that a job is a bundle of tasks. If AI takes over one or two, the person is still useful for the rest. If AI nibbles at every task, there is nowhere to shift to, and the firm needs fewer people. A companion finding suggests why the shifting is feasible. AI productivity gains show up when people apply skills they already have, and they vanish when people use AI to learn something new When does AI actually boost worker productivity?. Moving to the non-displaced tasks is the easy kind of move, because it uses existing skills. That link is my inference. The notes don't test it directly.

Firms don't respond to the same exposure in the same way. Firms with higher AI exposure replaced online-marketplace freelancers with AI tools faster and more cheaply than less-exposed firms did. That points to returns to scale in a firm's own AI capability, not a uniform spread of the technology Do firms substitute labor for AI at different rates?. That evidence is about contract labor, not employees, so it may not carry over. Even so, the same exposure numbers can produce different hiring outcomes depending on how ready a firm is to act on them.

Concentration also has a who-dimension, and here the picture is uneven. AI exposure clusters among high-skilled, high-paid workers in male-dominated occupations. In female-dominated occupations it spreads evenly across skill levels, so lower-paid, lower-skilled women face broad exposure with fewer resources to adapt Does AI exposure hit low-wage workers harder in some fields?. Under the task-level story, broadly spread exposure is the kind that leaves the least room to reallocate. The corpus doesn't test that pairing. Work that has actually been handed to AI in structured workflows also concentrates in information-intensive occupations and follows technical capability, not the routine-task predictions many people expected Where have workers actually delegated tasks to AI?. The broader review of generative AI and inequality says the direction of the effect depends on access, integration, and incentives, not the technology alone Does generative AI inevitably worsen or reduce inequality?.

The corpus leaves gaps. It doesn't say whether reallocated workers keep their pay or status, or how long the offset lasts as AI capabilities grow into the tasks workers moved to. It shows that the shape of exposure changes what firms do, but it doesn't show that the offset is permanent.