In a study of customer-support agents, an AI assistant raised productivity, helping novice and low-skilled workers most, with little effect on the most experienced. Knowing how such effects spread across the economy would show which workers gain and how their jobs change.
What we know
Early studies show large productivity gains on some tasks.
What blocks us
Most studies cover a few tasks or companies; the effects on the whole economy are not yet known.
Current leads
Measuring how AI is actually used at work, policies for transitions.
Sub-questions
Why do newer, less-skilled workers gain the most from AI assistants, while experts gain little?
Do workers who use AI assistants actually learn skills they keep?
Why does AI change productivity so differently from one worker to another?