Research

The AI Wage Gap: 30% More Per Hour

Based on Bick, Blandin, and Deming, NBER / Management Science, 2025

The NBER survey that tracks who uses AI at work also tracks what those workers earn. The 2025 numbers: workers who use generative AI daily earn about 30 percent more per hour than workers who do not use it at all. Weekly users earn about 15 percent more.

The gradient is the interesting part. More frequent use lines up with higher pay, step by step, which is what you would expect if the skill itself carries value.

An honest caveat, because that is the point of this page

This is a measured gap, and it is not proof of cause. Part of it runs the other way: people in higher-paid roles adopted the tools earlier, so some of that 30 percent was there before AI arrived. Anyone quoting this number as pure AI payoff is overselling it. We would rather show you the number with its limits attached.

Even with the caveat, two things hold. The gap appears across occupations rather than only in tech roles. And it scales with how often someone uses the tools, which points at skill rather than job title.

The floor is lower than it looks

Closing this gap does not require a degree or a bootcamp. The daily users in the survey are mostly ordinary professionals who worked the tools into normal tasks: drafting, summarising, checking, planning. Weeks of evening practice separate a non-user from a weekly user, and the wage data puts a number next to that step.

What this means for your career

Whether the gap is worth chasing depends on which of your tasks the tools can help with. For some roles that covers most of the week. For others it is a narrow slice, and pretending otherwise wastes your evenings.

That is what the quiz maps: your tasks against the measured data, so the answer rests on what you tell us about your week instead of a headline.