Will AI shrink the pay gap between weaker and stronger workers, even as it widens the gap between workers and capital owners?

This explores whether AI will narrow or widen pay gaps among workers in the near term. The corpus points to something easy to miss: there are two inequalities here, and AI seems to push them in opposite directions. Among workers, AI often compresses differences in performance. Between workers and owners of capital, it looks set to widen the gap.

The compression evidence is strong in controlled settings. In a 758-consultant experiment at BCG, the bottom-half performers gained the most from GPT-4, so the distance between weaker and stronger consultants shrank Does AI help lower-skilled workers catch up to top performers?. A randomized study of 1,174 adults found generative AI closed about three-quarters of the performance gap tied to education on a business problem-solving task. Less-educated participants also kept part of the gain after the AI was taken away Can AI narrow the education performance gap?. Acemoglu's task-based model fits this picture: AI's wage effects spread more evenly across demographic groups than earlier waves of automation did Will AI increase inequality less than past automation?. There's a catch, though. The BCG gains came when AI use was mandatory. When people choose whether to use it, the effect can reverse, because people who are already strong tend to adopt first.

The same Acemoglu model also predicts that capital income pulls ahead of labor income overall. Anthropic's scenario modeling goes further. As AI growth speeds up, the share of income going to labor falls, and knowledge-worker wages stagnate or decline even while GDP and average wages rise Does AI growth inevitably shift wealth away from workers?. At the far end, one AGI-economy model has human wages sliding toward the cost of the compute needed to replace them, with labor's share approaching zero What happens to human wages in an AGI economy?. Firm-level data hints at how this concentration starts. Firms with more AI exposure replace freelance workers faster and more cheaply, which suggests the returns go to companies that build AI capability in-house, not to everyone evenly Do firms substitute labor for AI at different rates?.

Who gets hurt in the near term is also uneven in ways a single average hides. Payroll data through mid-2026 shows no economy-wide job losses from AI. Young workers in AI-exposed occupations, however, are being hired at 19% lower rates, while experienced workers show no such gap Is generative AI displacing workers at economy-wide scale?. That's a less visible kind of inequality: the entry rung on the career ladder thins out. Exposure also splits by gender. In male-dominated fields it mostly hits high-paid workers. In female-dominated fields it spreads across all pay levels, including lower-paid women who have less room to adapt Does AI exposure hit low-wage workers harder in some fields?. How tasks are exposed matters too. When AI touches only a few tasks in a job, workers can shift to the rest and losses stay modest Does concentrated AI exposure enable workers to adapt and reallocate?.

So the corpus's honest answer is: AI may narrow skill-based pay gaps inside jobs while growing the gap between labor and capital. Which effect wins depends on choices more than on the technology itself. An interdisciplinary review finds that access, how AI gets built into work, and incentives decide whether it reduces or worsens inequality Does generative AI inevitably worsen or reduce inequality?. The corpus has more modeling and lab experiments than real-world wage data, so firm near-term predictions aren't yet well supported.