Building economic datasets now costs hours, not months, thanks to AI agents

AI agents can now build complex economic datasets from public sources, making a costly research task much cheaper and faster. Researchers can now create detailed datasets that used to take months of human labor for the cost of a few hours of AI time.

For decades, building large, clean economic datasets was a huge bottleneck, requiring armies of research assistants. This paper shows that AI can automate much of that grunt work. This means economists can ask and answer more complex questions, or update existing datasets more frequently, without the massive upfront cost.