Complements, Substitutes, AI and the Way Forward

One of the most popular blogposts on this blog is one that I wrote over five years ago: a simple explainer post about complements and substitutes.

It’s part of the arsenal of an economist, an understanding of the concept of substitutes and complements, and it is useful in many surprising and unexpected ways. But never has its use been as important as it is in understanding the importance, the threat and the advantages of AI. A video that I have often linked to in the past, and will probably link to many times again helps make this point clear:

When Steve Jobs says computers are like bicycles for the mind, he is saying that our mind becomes more powerful when we work with computers, rather than instead of them (substitutes) or infinitely worse, without them (almost all examinations conducted in higher education in India today).

And if you want to think about your career in this brave new world of ours, you really should be thinking about working with computers. Not against, or without. As it turns out, this is surprisingly hard to do for most of us. I invite you to walk into a higher education institute of your choice and listen to professors talk about how many students are copying during examinations. Nobody seems to ask why it is right and appropriate to check how good students are at doing work without computers. Why is this a skill that we’re building for folks who will be working in the 21st century?

And if you are learning how to work “effectively” without a computer – and again, that is what we train you for when we make you write three hour pen-and-paper examinations in higher education – you are destroying your ability to earn more in the future.

I’m being quite serious.

The key questions will be: Are you good at working with intelligent machines or not? Are your skills a complement to the skills of the computer, or is the computer doing better without you? Worst of all, are you competing against the computer?

Cowen, Tyler. Average is over: Powering America beyond the age of the great stagnation. Penguin, 2013.

A lot of people are scared about job losses as a consequence of the rapid development of AI, and with good reason. AI can today do quite a few jobs better than humans can, and more than its current capabilities, what keeps a lot of us up at night is the rate of improvement. Not only is AI very good already, but it is noticeably better than it was last year. And for the pessimists among us, the scarier part is that not only will AI be even better next year, but the rate of improvement will also improve. That is, the improvement in AI’s abilities will not only be more in 2023 compared to 2022, but the difference between 2023 and 2022 will be higher than was the difference in 2022 compared to 2021. And that will be true(er) for 2025, and for 2026 and, well, there’s no telling where we’re headed.

But this is exactly why studying economics helps! Because both Steve Jobs and Tyler Cowen are, in effect, saying the same thing: so long as you plan your career by using computers/AI as a complement, you’re going to be just fine. If you think of your job as being substitutable – or if your job is, or will be, substitutable by a computer – well then, yes, you do have problems.

An underappreciated point is the inherent dynamism of this problem. While your job may not yet be a substitute for AI, that is no reason to assume that it will not be substitutable forever:


For example: is Coursera for Campus a complement to my teaching or a substitute for it? There are many factors that will decide the answer to this question, including quality, price and convenience among others, and complementarity today may well end up being substitutability tomorrow. If this isn’t clear, think about it this way: cars and drivers were complementary goods for decades, but today, is a self-driving car a complement or a substitute where a driver is concerned?

https://econforeverybody.com/2022/04/18/supply-and-demand-complements-and-substitutes-and-dalle-e-2/

But even so, I find myself being more optimistic about AI, and how it can make us more productive. I haven’t come across a better explainer than the one that Ethan Mollick wrote about in a lovely post called Four Paths to the Revelation:

I think the world is divided into two types of people: those obsessed with what creative AI means for their work & future and those who haven’t really tried creative AI yet. To be clear, a lot of people in the second category have technically tried AI systems and thought they were amusing, but not useful. It is easy to be decieved, because we naturally tend try out AI in a way that highlights their weaknesses, not their strengths.
My goal in this post is to give you four experiments you can do, in less than 10 minutes each, with the free ChatGPT, in order to understand why you should care about it.

https://oneusefulthing.substack.com/p/four-paths-to-the-revelation

All four examples in this post are fantastic, but the third one is particularly relevant here. Ethan Mollick walks us through how AI can:

  1. Give you ideas about what kind of business you might be able to set up given your skills
  2. Refines a particular idea that you would like to explore in greater detail
  3. Gives you next steps in terms of actualyl taking that idea forward
  4. And even writes out a letter that you might want to send out to potential business collaboarators

His earlier posts on his blog also help you understand how he himself is using ChatGPT3 in his daily workflow. He is a professor, and he helps you understand what a “mechanical” professor might be able to do

To demonstrate why I think this is the case, I wanted to see how much of my work an AI could do right now. And I think the results will surprise you. While not nearly as good as a human professor at any task (please note, school administrators), and with some clear weaknesses, it can do a shocking amount right now. But, rather than be scared of AI, we should think about how these systems provide us an opportunity to help extend our own capabilities

https://oneusefulthing.substack.com/p/the-mechanical-professor (emphasis added)

Note the same idea being used here – it really is all about compementarity and substitutability.

AI can already create a syllabus and refine it; it can create an assignment and refine it; it can create a rubric for this assignment; it can create lecture notes; and it can write a rap song about a business management concept to make the content more interesting for students. I loathe the time spent in creating documentation around education (every single teacher does) and it would take me a long time to come up with even a halfway possible rap song about substitutes and complements.

That last statement is no longer true: it took me twenty seconds.

Here are examples from outside the field of academia:

The question to ask isn’t “how long before I’m replaced?. The question to ask is “what can I do with the time that AI has saved me?”. And the answer to that question should show that you are thinking deeply about how you can use (and continue to use!) AI as a useful complement.

If you don’t think about this, then yes, I do think that you and your job are in trouble. Get thinking!

Leave a Reply