The Jevons Paradox means good news for tech jobs in the age of AI

Are you buying the doomsayers' jeremiads? Here's a counter-narrative.

You're probably sick of reading all the bovine leavings about how AI is going to make information workers obsolete. It's coming from every angle:

• the hysterical workers themselves, convinced that unemployment is a permanent institution rather than a cyclical function

• the UBI boosters who have been waiting for this moment their whole life

• and, of course, the AI bros themselves, who were never told no as children, and probably won lots of participation trophies, so they have stage IV main character syndrome

So, if you find yourself believing the hype, JUST STOP. Let me reintroduce some sense into the discussion.

---

---

Are you familiar with the Jevons Paradox? In short, it goes like this:

Any technology that lets you get more benefit from a smaller amount of a given resource will cause greater usage of that resource.

This phenomenon occurs in all sorts of places.

• More fuel efficient cars don't result in less oil being used, because people just end up driving even more, resulting in more oil being used.

• Data compression doesn't result in less server space or bandwidth being used, because now it's possible to stream fast media for cheap, so data usage goes through the roof.

• Modern farming technology that produces more food per acre doesn't mean we have less farmland on earth, because food becomes cheaper meaning the population grows, and with it the demand for food.

Often this paradox is framed in negative, even pessimistic, terms like "why bother making better technology if it won't lower the use of resources". Basically a bunch of Luddite whining. But there is a flip side to this paradox that nobody is discussing.

Human labor is a resource

In fact, human labor is a resource that AI just made much more productive. And it's only getting better. In each of the above examples of physical resources, the cost per unit starts out high enough that people only use them for the most crucial stuff. With the price of oil, if each mile driven by car is expensive, then you'll only drive for things where walking, a bicycle, or public transit won't cut it. As it becomes cheaper, the convenience of driving starts to outweigh the cost. And once it's cheap enough, you'll just start driving for fun without really thinking about the cost.

With human resources, every employee carries a fixed overhead cost in addition to the actual expense of their labor. No matter how many hours someone works, simply keeping them on your payroll and maintaining a legal liability for them costs you money: insurance policies, HR staff, equipment, and managerial complexity. And it scales exponentially. Managing 1000 people is not 10x more expensive than managing 100 people. It's way more than that. The result of all this is that, at a certain point, it's just not worth hiring more employees. If you pay someone $100,000 per year after taxes, you have to get far more than that in returns from them to make the hire worth it. Even if that person produces $200,000 in value (and guaranteed, crude corporate accounting is only probably catching $110,000 of that recouped value), it might not be worth it.

But as AI enables a single person to produce more value than without AI, the calculus changes. Most people (the AI douchebags most of all, ironically) vastly underestimate the potential of a single person to create value. The things holding their potential back include bad management, information asymmetry, and clunky tools. Even when the employee themselves is kind of slacking, one has to ask why they are slacking? Did bad tools and the lack of access to information break their flow? Well, AI could help prevent all that.

So why hasn’t it happened yet?

It's naive to think that this principle doesn't apply to people anymore than it does to oil, floating operations per second, or farmland, and yet it never gets brought up. One reason is that the Jevons Paradox always has a bit of a lag, but in the case of knowledge jobs, that lag is causing hardship for people, and thus the lag seems to be interminable. But there may be a longer delay than usual before the paradox kicks in, and this is why:

People come with overhead. As I said earlier, an employer doesn't just take on the cost of an employee's paycheck. They create more work for whomever the employee reports to, they need physical equipment if it's onsite and software even when it’s remote, and that needs to be both purchased and tracked. You have to file their paperwork with the government. And then you actually have to find the employee, and that isn't always easy since humans are infamously non-standardized. And there's that chance that you might need to let them go, and that will come with its own set of headaches, even more so in Europe.

Combine this with the fact that the increased value that an employee provides due to AI augmentation might not be known immediately to the company. Let's be real. Companies drastically undervalue their employees given the shortsighted and downright infantile way the tech companies laid off workers in their thousands the last few years. So to assume they'll immediately pick up on the increased value of AI-augmented employees is equally naive.

Things will be fine

But that doesn't mean that some company won't figure it out. And when they do, they are going to start pummeling the competition. And you can bet that competition will start copying the winning company with the same mindlessness as they copied each other's layoffs.

The big question is which sectors of the market are going to move first on this advantage? Will big companies start hiring more people because they already have the capacity to take them on? Or will smaller startups use this as an opportunity to ask for more money from VCs to bring on the newly valuable employees? I think the first place we'll see the boom will be in the freelancing and consulting industries, since the risk for employers will be lower.

So there it is. Some good news in a time when the press is hell-bent on telling you the sky is falling. As for how AI will increase the value of employees, that's a subject for a future article. For now, go play outside. It's springtime.

Dogs are forever in the push-up position.

---