Does AI spreading through jobs squeeze paychecks, or do workers just shift to the tasks it can't touch?

This explores what happens to pay as AI spreads from a few specialized jobs into many kinds of work. The short answer from the corpus is that how the spread happens matters as much as how far it goes. AI doesn't wash evenly over the labor market. It concentrates in particular tasks, firms and kinds of workers, and that unevenness decides who gets squeezed. One analysis of firms from 2010 to 2023 found that higher average AI exposure does reduce demand for labor. But when exposure is concentrated in just a few tasks within a job, workers can shift toward the tasks AI doesn't touch, and the net employment effect stays modest Does concentrated AI exposure enable workers to adapt and reallocate?. The danger comes when AI capability covers most of what a job involves, leaving nowhere to move.

The early evidence shows the spread follows lines that already exist. Job postings across ten countries show AI skill demand clustering around a technical core (Python, SQL, machine learning) inside STEM occupations, while non-technical jobs move away from that core instead of toward it Is AI creating common skills across jobs or deepening divisions?. Where workers have actually handed tasks over to AI, it's mostly in information-heavy work. That pattern follows what the technology can do, not the old prediction that routine jobs go first, and the wage pattern flips at the advanced-degree level Where have workers actually delegated tasks to AI?. Gender adds another split. In male-dominated fields, exposure falls mostly on high-paid, high-skilled workers. In female-dominated fields it's spread across every skill level, so lower-paid women end up exposed with fewer resources to adapt Does AI exposure hit low-wage workers harder in some fields?.

Two more findings suggest the gains won't spread evenly either. Firms that are already heavily exposed to AI replace freelance workers with AI tools faster and more cheaply than other firms, which points to returns to scale in building AI capability in-house, not a uniform wave of adoption Do firms substitute labor for AI at different rates?. And the well-known productivity boosts from AI show up when people apply skills they already have. When workers used AI to learn something new, the gains disappeared and their learning suffered When does AI actually boost worker productivity?. Taken together, AI looks more likely to raise the value of existing expertise than to help workers climb into better-paid roles.

The long-run scenarios go further. Anthropic's scenario modeling finds that as AI growth speeds up, labor's share of income falls and capital's share rises. Average wages can still go up while knowledge-worker wages stall or fall, because ownership is concentrated and switching occupations is hard Does AI growth inevitably shift wealth away from workers?. A more radical analysis argues that in a full AGI economy, human wages stop tracking the value of the work and settle at the cost of the compute needed to replicate it. Labor's share of GDP heads toward zero even if some work stays human What happens to human wages in an AGI economy?. On timing, though: agents that win benchmark contests still fail at long, real professional workflows, so capability on paper is running well ahead of capability on the job Why do agent benchmarks not predict real economic value?.

The part you may not have expected to care about is that wages are also a source of leverage, not just income. One line of work argues that society's institutions stay roughly aligned with human interests partly because they depend on human workers who care about outcomes. As AI replaces that labor bit by bit, that quiet check weakens, and institutions can drift away from what people want in ways that may be hard to reverse Does incremental AI replacement erode human influence over society?. So the wage question is partly about whether human work keeps any say in how the economy runs.