Rogé Karma

I think the strongest case is that the broad educational trends we are experiencing are sort of exactly what you'd expect if AI was the culprit of displacing workers.

So, right, I mentioned that since 2022, and this is actually really, if you look at since November of 2022, when ChatGPT was released, again, the unemployment rate for young college graduates has risen nearly twice as fast as the rest of the workforce.

That single statistic is exactly what you might expect if AI was displacing entry-level white-collar workers.

What is the kind of work that recent college graduates tend to do?

It is things like making PowerPoint slides.

It's working in Excel.

It's writing first drafts of reports.

It's doing a lot of the sort of text-based entry-level work that we already know that AI systems like Claude or ChatGPT can do pretty well.

Another big one is writing a lot of first draft of code.

And so you look at these things, you look at the kinds of tasks that AI is already good at, then you look at the type of workers who tend to perform those tasks, you see all of a sudden,

Those workers increasing in unemployment faster than they normally are.

You put that all together, it seems like an AI story.

Derek, let me tell you, economists normally don't agree on much, but this is an issue where there's especially vehement and sometimes quite spicy disagreement.

There are economists all over the gamut here, right?

There are ones who think that, and papers that seem to find that

AI is already displacing a lot of entry-level, for example, software engineers.

Then on the other side, there are economists who think that it has had basically no labor market impact at all, lots of views in between, and really no clear consensus right now.

What is the nub of their disagreement?

I'm going to get a little nerdy here.

One of the main focal points of disagreement, I would say the nub, concerns this statistic called the unemployment rate.