A blog post from BEA Director Vipin Arora
I’m sure many of us remember playing the “don’t say it” game. Someone picks a word to avoid, and the first person to utter that word loses. Can you imagine how quickly the game would end these days if the word was AI?
Statistical agencies are no exception. Artificial intelligence is all the rage in our world, both how it’s being used within our organizations and how we’re measuring it. Putting aside our own use (more soon on this), my sense is that there are three broad aspects of measuring AI that statistical agencies are grappling with.
Estimating the impact of AI on workers, the nature of work, productivity, and the economy is one aspect that is top of mind for many people. There is also tremendous interest in understanding where, how, and how much AI is used. We’re actively improving measurement in both these areas across U.S. statistical agencies.
At BEA, we’ve also made significant progress in a third area: estimating how much AI production—through data center construction, algorithm development, energy use, and other activities—contributes to U.S. economic activity.
I’m excited to let you know we’ll be releasing groundbreaking experimental statistics on the size of the American AI economy this year. While these are early estimates, they’re a big step towards improving AI measurement and broadening our understanding of its impacts.