People often forget that the laws of economics aren't unchangeable like the laws of physics. Economics is influenced by how people view the world at any given time, which is why economic theory sometimes shifts over time. It's why you can study the laws of supply and demand in Macroeconomics and then can't understand why Giffen Goods exist. It's why "modern monetary theory" is suddenly in vogue, despite contradicting much of the established economic orthodoxy. Much of the changes to economic theory from the last wave of tech innovation are captured in a great book called Capitalism Without Capital, which I highly recommend.
The question I want to address briefly this week is - what changes might AI bring that could affect the fundamental laws of economics? I'm going to take a stab at three ideas.
First, knowledge curves are going to factor in to economics. Given that AI can learn much faster than humans, and what it can learn is tied to the various data representations it has access to, and the size of those data sets, this needs to be added to economics. One place it could have a huge impact is on how knowledge interacts with supply curves. In a manufacturing world, supply curves are tied to the cost to manufacture as the scale of that manufacturing changes. After Web 1.0, supply curves changed because distributing knowledge once it is produced had a near zero marginal cost. Now AI is going to change knowledge production. In a knowlege production world, shapes of the curves are tied to the cost to produce knowledge. And various types of knowledge could have different costs to produce. Expect these to interact with supply curves in interesting ways.
Second, the concept of a demand curve could change if AI gets so good at pressing our buttons that we can be sold nearly anything. Imagine an AI that is so good it can figure out the language and the process to sell ice to eskimos. Instead of a demand curve where there is some latent market demand for a good depending on the price, there is now almost some type of "push" curve, where, depending on how much work the AI wants to put in, it can almost always convince you to buy. Rather than fulfilling demand, it becomes about pushing a product on someone because you know how to push their buttons. So how much work is it to sell a certain level of something? There could be a curve for that. This is a long way off, but, we could very well get there.
Third, AI could introduce data driven discontinuities into the mathematics of economics at many points. Why? Because what AI can do depends on the data it has. Smooth curves for common economic concepts could turn into step functions as marginal amounts of data suddenly push models to new levels of performance in discrete jumps.
The laws of economics aren't fixed forever. As AI impacts society in many ways, and as AI makes more decisions for humans that we no longer make ourselves, it is very possible the core principles of both macro and micro economics shift. If you have ideas for how that may happen I'd love to hear them.
Thanks for reading.