Our last piece — Our Only Plan is to Grow Up — was on the expansive side.

It covered:

In the spirit of May-cember (“a month as busy as December, but with no presents for anyone involved”), let’s keep this one short… Hopefully, you’ll go back to what feels like the best use of your time from the above list too.

In the meantime, I think my last conversation on TREUSSARD TALKS, with Andrea Eisfeldt, is a very special one.

I suspect you’ll agree.

Dr. Eisfeldt in three bullets:

And I should have introduced Andrea as “Madam Vice-President,” as she is the Vice-President of the American Finance Association this year—the leading body in academic finance in the U.S., which publishes The Journal of Finance.

Need I say more?

When you listen, you’ll hear the following.

We talked about her seminal academic work on the rise of intangible assets (valuable non-physical assets, like employee know-how or corporate brand) on company balance sheets.

We also talked about how we have undercounted the total compensation going to workers recently, given the rise of equity compensation over the last few decades (particularly in tech and finance, and in startup-land at large).

In short:

→ Squishy “human capital” has become a bigger part of what makes a firm valuable (as opposed to “plant and equipment” in the industrial days).

→ In turn, corporate equity has become a larger part of how valued labor has been compensated for its trouble.

Funny how that all makes sense when you see both sides of it—the way Andrea does.

This is how systems evolve.

Nothing happens in a vacuum.

Things are interrelated, and one change over here will drive a change over there.

Andrea has done some of the most insightful work I’ve seen so far on the impact of AI on financial markets (forthcoming in The Journal of Finance) by taking a critical detour into the labor market.

Andrea and her colleagues went straight to the Bureau of Labor Statistics data.

It turns out we officially have 900 occupations in America.

If that’s not reductive enough for you, each occupation is assigned 20 tasks by government statisticians.

So if you have a job, you’re responsible for 20 tasks (some deemed “core” and some “supplemental”). That’s it…

You can thank the economists at the BLS for making you feel good about yourself.

What Andrea and her coauthors did was estimate how much of these tasks could be accomplished by AI, especially the core tasks.

Projected onto stock returns post-ChatGPT, it looks like the market is making a large and statistically significant bet that either AI will make each of these workers a corporate superhuman or they’ll get canned (thereby reducing costs to their firms).

Given how much our economy depends on consumers consuming out of labor income, we should hope we’re getting a lot of corporate, AI-empowered superhuman colleagues. Otherwise, based on my own analysis combining BLS data with Andrea et al.’s data, we have a couple of trillion dollars of wages on the line… It would not be good for the economy if those went away, even if we’re only talking about a short-term bout of unemployment while the labor force reshuffles to align with this new world.

But here is possibly the most critical thing that came out of my conversation with Andrea.

If you buy into the idea that AI can create near-infinite productivity gains, the only reason things would get stuck and not get done is because someone somewhere needs to tell the machines what to do, how to do it, and assess the quality of the resulting “work.”

In other words, until the machines can replace all subject experts and all bosses—and just manage things among themselves—humans (those with the real knowledge, leadership, and vision) become more valuable, not less.

This is from Andrea on TREUSSARD TALKS:

“That bottleneck in production becomes extremely valuable and constrains growth.

It's like if we're trying to pour water out of a bottle, how fast is it going to come out? It doesn't matter how wide the top is; it's how narrow the bottom is.

And so the kind of humanity that we still need in order to prompt AI, to oversee AI, to manage it, to decide what task to assign it to—that's all going to become a constraint in terms of how fast we can grow and what we can do. But it's also going to become a highly compensated task.

As for distributional effects, it's clearly gonna have distributional effects. And I have some of the same concerns that you do. It's kind of hard for everybody to adjust to it, and some bad things may happen when you do that.

This is a very fast-progressing technology that's also very fast-diffusing. So I agree with you that we have to get ahead of what kind of fallout we might see.”

At the risk of sounding like an economist…

On the one hand, it’s a pretty solid case for humanity.

On the other hand, if we’re going to have major winners and major losers in this transition, we’d better have more than just humanity.

We’re going to need a boatload of empathy for each other to see this through together.

Listen to my conversation with Andrea Eisfeldt: