This work wouldn’t have been possible without my fantastic coauthors Ruth Appel*, Peter McCrory*, Michael Stern, Miles McCain, Tyler Neylon, and and the enormously talented team behind the project. Stay tuned for more!

See the report and interactive website for much more, including a job explorer where you can see what fraction of tasks in each job people are using AI for.

www.anthropic.com/economic-ind...

Finally, in “power laws rule everything around me,” it turns out that usage across different use cases is, of course, distributed in the same way.

Unsurprisingly, we see much more automation on the API, reflecting how the API tends to use be used more programmatically rather than through an interactive interfaces.

We also release first-of-its-kind insights on how people are using the Anthropic 1P API, using our privacy-preserving analysis system.

Use cases range from interior design to analyzing financial reports to helping with freight logistics.

We also see a surprising correlation between usage and interaction style. People in higher-use countries tend to use Claude more collaboratively rather than simply delegating to it.

We’re not sure why this is, and would be interested to see more research on here.

Second, AI use is dramatically uneven across the world. We see a strong correlation of usage per capita with GDP per capita, raising the prospect of a new Great Divergence.
en.wikipedia.org/wiki/Great_D...

First, users are delegating more tasks to Claude.

This could reflect both increased user trust in AI and model improvements.

We see more activity associated with code creation and less with debugging, which might mean users can accomplish their goals more in a single shot.

Links first, then some takeaways:

Interactive website: anthropic.com/economic-index
Report & data: www.anthropic.com/research/ant...

"gets rave reviews from clients"

Thanks to everyone else who gave feedback and helped out along the way.

There's lots more to come, so stay tuned.

6/6🌅

This work was a huge group effort.

Thanks to my amazing coauthors, including co-lead Kunal Handa, Miles McCain, Saffron Huang, Esin Durmus, Sarah Heck, Jared Mueller, Jerry Hong, Stuart Ritchie, Tim Belonax, Kevin K. Troy, Dario Amodei, Jared Kaplan, Jack Clark, and Deep Ganguli

5/

Third, there are many limitations to our work! (See Section 4.1 in the paper). We're working on making progress on them.

We’re also excited to see what others do with our data—please get in touch via this form with any feedback or input:

docs.google.com/forms/d/e/1F...

4/

Second, because our methods are fully automated, we can run the same process over time on a recurring basis.

This gives us a moving picture of how AI is advancing across the economy, and identifies leading indicators that we can use to plan.

3/

While those links have all the details, I wanted to call out two additional points:

First, I’m excited that our methods give insight into *how* work is done with AI

There’s a lot of interest in *what* tasks AI will do, but *how* AI changes the nature of work is also crucial.

2/