When AI can do parts of your business's work, do companies lay people off — or change what those people do?

This explores whether high-exposure firms cut jobs or reshape roles. The corpus points to both, and the deciding factor is how the exposure is shaped, not how much of it there is. Across firms from 2010 to 2023, higher average exposure does reduce labor demand. But when the exposure is concentrated in a few tasks, workers can shift to the tasks AI doesn't touch, and the net employment effect ends up modest Does concentrated AI exposure enable workers to adapt and reallocate?. A job where AI can do one slice of the work behaves very differently from one where it can do a bit of everything.

Where firms do substitute, they don't all do it at the same pace. Firms with higher AI exposure replace online freelance-marketplace workers with AI tools faster and more cheaply than less-exposed firms do. That looks like a payoff to building in-house AI capability, not a uniform wave of technology spreading through the economy Do firms substitute labor for AI at different rates?. Note what was measured, though: contract workers being swapped out, not employees being laid off.

The reshaping side shows up in where workers have actually handed tasks to AI. That is mostly information-intensive occupations, and it follows what the technology can do more than the old routine-versus-non-routine automation story would predict Where have workers actually delegated tasks to AI?. Inside those jobs, interviews with workers show a quiet change in what the work is. People hold on to cues like their voice and where the material came from. The effort, attention and uncertainty behind an AI-assisted deliverable disappear into the finished product, because output-centered work treats the finished task as proof that the work happened Which workplace cues survive AI mediation and which disappear?. A role can look unchanged from the outside while its content has shifted toward producing and vouching for output.

Who gets reshaped, and who gets displaced, is uneven. Exposure concentrates among high-skilled, high-paid workers in male-dominated occupations. In female-dominated ones it spreads evenly across skill levels, so lower-paid, lower-skilled women face more exposure with fewer resources to adapt Does AI exposure hit low-wage workers harder in some fields?. Job postings from ten countries show reshaping isn't a shared upskilling either. AI skill demand clusters around Python, SQL, machine learning and data analysis in STEM roles, while non-technical occupations move away from that core, not toward it Is AI creating common skills across jobs or deepening divisions?.

Treat exposure numbers as an upper bound on near-term displacement. An analysis of 960 real occupational workflows found agents do well on abstract contests but fail at long-horizon professional tasks, because the field has measured contests, not work Why do agent benchmarks not predict real economic value?. The corpus doesn't include firm-level layoff studies, so the fair reading is this: substitution is real but concentrated in outsourced, easily separated work, and inside firms the more common change is a redistribution of tasks, with the costs falling unevenly.