Eva de Francisco

I just finished reading “The worlds I see” by Dr. Fei-Fei Li, a fascinating autobiography and story about the origins of modern AI. Back in the day, scientists tried to focus on teaching computers how to do things using algorithms that followed rational and logical rules, but with this strategy led them nowhere! Too many rules, different rules for different problems, exceptions, these algorithms could only solve very simple problems, and answered very specific questions in closed or controlled environments.
Then, after quite a few years, the focus switched to letting computers learn the way we actually learn, this is, by being exposed to life, this is, many different images, things, situations, experiences, and by finding patterns and associations and making decisions without being able to fully explain why.
This is a big blow for most economists in academia, myself included. Back in economic theory class, we learn that “agents”, yes, we want to be taken seriously and use words like “agents” instead of just “people”, like variety and get satiated if they get too much of the same thing. So for example, if I have a blue and a yellow t-shirt, and I am presented with the opportunity to buy a red or a yellow t-shirt, then, economic theory says I will buy the red one, pretty much disregarding that I may just love how great yellow looks on me and want ten slightly different yellow t-shirts. This example is silly but many others are not, and hopefully you get the point.
Yes, some of you may know that behavioral economics had started to address our “lack” of rationality when taking some decisions, like for example when people choose their retirement saving portfolio. Most of us tend to use a rule of thumb for choosing, this is, if we are offered two sets of investment options with very different risk profiles, we pick to invest half of our savings in each, no matter the actual risk offered in each of them.
Macroeconomics, my field, the study of the economic decisions of big groups (government, private sector, or consumers), and the interactions and policies that affect their decisions, always have had a hard time to begin with, even before the arrival of AI and big data. As you can imagine, it is impossible to isolate changes and conditions outside a tube lab, but some of us think it is still worth to try to understand the broad implications and problems of the sum of our collective decisions. “Should we invest 1% or 2% of the government budget in education?” “What type of taxes should we use for this?”
To give the field some discipline, we ask the agents in our big groups to follow and behave by the economic axioms we learnt in economic theory (more is always preferred to less, we like variety, we understand how we make decisions and what makes us happy, and we make decisions to reach our goals, to name the most basic ones) and see if our models resemble the big trends we see in the data.
Microeconomics, always took on to ask more well-defined and specific questions, such as “By how much house prices fall when a prison opens nearby?” As of lately, with more and more data available, the profession has placed a big emphasis on the idea of “identification”, this is, being able to pinpoint exactly why we are observing something, to prove causality between A and B.
This trend is narrowing the questions economists ask themselves making them less interesting and more irrelevant by the day, so economists keep losing ground to data scientists and AI experts.
Economists going against this trend are having a hard time publishing in academic journals and are leaving academia to pursue other endeavors, and maybe that will be for the better, we don’t know yet.
Today, to keep on publishing, papers have to be longer and longer, more technical, more difficult to read, and just plain more boring, with more robustness’ checks, and additional online appendixes. The effect is that less people read our work and our impact diminishes.
In my opinion, this creates very few winners, and probably not for long, but the real losers are all of us and my beloved profession and science, Economics.

I just finished reading “The worlds I see” by Dr. Fei-Fei Li, a fascinating autobiography and story about the origins of modern AI. Back in the day, scientists tried to focus on teaching computers how to do things using algorithms that followed rational and logical rules, but with this strategy led them nowhere! Too many rules, different rules for different problems, exceptions, these algorithms could only solve very simple problems, and answered very specific questions in closed or controlled environments.
Then, after quite a few years, the focus switched to letting computers learn the way we actually learn, this is, by being exposed to life, this is, many different images, things, situations, experiences, and by finding patterns and associations and making decisions without being able to fully explain why.
This is a big blow for most economists in academia, myself included. Back in economic theory class, we learn that “agents”, yes, we want to be taken seriously and use words like “agents” instead of just “people”, like variety and get satiated if they get too much of the same thing. So for example, if I have a blue and a yellow t-shirt, and I am presented with the opportunity to buy a red or a yellow t-shirt, then, economic theory says I will buy the red one, pretty much disregarding that I may just love how great yellow looks on me and want ten slightly different yellow t-shirts. This example is silly but many others are not, and hopefully you get the point.
Yes, some of you may know that behavioral economics had started to address our “lack” of rationality when taking some decisions, like for example when people choose their retirement saving portfolio. Most of us tend to use a rule of thumb for choosing, this is, if we are offered two sets of investment options with very different risk profiles, we pick to invest half of our savings in each, no matter the actual risk offered in each of them.
Macroeconomics, my field, the study of the economic decisions of big groups (government, private sector, or consumers), and the interactions and policies that affect their decisions, always have had a hard time to begin with, even before the arrival of AI and big data. As you can imagine, it is impossible to isolate changes and conditions outside a tube lab, but some of us think it is still worth to try to understand the broad implications and problems of the sum of our collective decisions. “Should we invest 1% or 2% of the government budget in education?” “What type of taxes should we use for this?”
To give the field some discipline, we ask the agents in our big groups to follow and behave by the economic axioms we learnt in economic theory (more is always preferred to less, we like variety, we understand how we make decisions and what makes us happy, and we make decisions to reach our goals, to name the most basic ones) and see if our models resemble the big trends we see in the data.
Microeconomics, specially nowadays, with more and more data available, has taken on to ask more well-defined and specific questions with a big emphasis on the idea of “identification”, this is, being able to pinpoint exactly why we are observing something specific, like “By how much house prices fall when a prison opens nearby?”
This trend is narrowing the questions economists ask themselves making them less interesting and more irrelevant by the day, so economists keep losing ground to data scientists and AI experts.
Economists going against this trend are having a hard time publishing in academic journals and are leaving academia to pursue other endeavors, and maybe that will be for the better, we don’t know yet.
To keep on publishing, papers have gotten longer and longer, more technical, more difficult to read, and just plain more boring, with more robustness’ checks, and an additional online appendix. The effect is that less people read them and their impact diminishes.
In my opinion, this creates very few winners, and probably not for long, but the real losers are all of us and my beloved profession and science, Economics.