Peter McCrory
π€ SpeakerAppearances Over Time
Podcast Appearances
Yeah, so this is the labor impact report that I was alluding to before.
I think the motivation here was trying to figure out how do the capabilities that we see in all of the benchmarks and that we're tracking very closely over time, how do those model capabilities meet the real world?
As we've just mentioned, there are reasons why it's easier or harder to adopt AI in profession A versus profession B. Software engineers are much more likely to use Claude than other knowledge workers, for example.
Our motivation when producing the observed exposure measure was, where is Claude being used in automated ways for work purposes?
Because that type of adoption is likely to signal
or at least has the greatest chance of predicting displacement.
So in our analysis, we classify usage between automated use, so you give a simple directive to Claude and Claude completes it, or augmented use, where you go back and forth.
You work with Claude more like a thought partner.
If you're using Claude more as a thought partner,
That clearly points in the direction of labor augmentation.
That tends to be a source of pushing up demand for labor and expertise.
Whereas if Claude is able to complete the task end to end in a fully autonomous and reliable way,
You can think about data entry work where you're grabbing bits of information from documents and compiling it into a different database.
Increasingly large language models can just do that task maybe even better than humans.
We would expect that those are the jobs where demand
for those workers is likely to subside as businesses ultimately adopt.
Interestingly, we do see that occupations that have high AI exposure, computer programmers, data entry workers, as I mentioned,
These jobs are also expected by the Bureau of Labor Statistics in the U.S.
to have a weaker job growth over the next decade.
So it does seem like we're picking up some structural dimension of where the economy is headed, and we think that this notion can help us track whether there's any widespread displacement today.