AI and the Undergraduate Economics Major

 I am on academic leave from USC for the entire 2026-2027 academic year. When I return to USC, I will teach only undergraduate classes during the 2027-2028 academic year. After reading this recent USC Trojan piece on AI and the Undergraduate experience, I want to share some thoughts on how I will structure my classes.

I encourage you to read my paper; “Will AI Improve Undergraduate Economics Education?”. My paper is a highly optimistic paper. Pessimists shouldn’t bother to read this.

I fed my paper to Chat GPT and asked to list 12 interesting points I make in this paper and to rank them from most to least interesting. I will discuss some of these below.

In my humble opinion, universities have been really lazy in having the professor give a lecture where the highly heterogeneous student body sits there. Advanced students are bored, mediocre students are confused, and stoned students are spaced out and not paying attention. This “mismatch” can be solved by AI.

Everyone has a passion and side-hustle and Econ helps everyone think things through. Suppose a young person wants to understand what the Federal Reserve Chair really does and how this impacts the daily life of young people.

In my article, I discuss a type of “Flight simulator” where the AI enabled student is placed into Kevin Warsh’s seat and must make a decision facing uncertainty. I am explicit about what information Warsh has and doesn’t have and what are the consequences of his decisions.

This “gameification” of the Fed’s job is the right way to teach intermediate macro. One can then use basic discrete math to teach students dynamic programming and the permanent income hypothesis of consumption and firm investment decisions (where to locate a new factory and how much to produce) as a function of factor prices, taxes, regulations and expectations.

Advanced students (and USC has some great young minds) can be pushed much faster into advanced material using AI. With some undergraduates, I give them my recent papers and tell them to read them. With AI, they can ask “dumb questions” about my methods used in my statistical analysis and with a little bit of effort they can build up their intuition about how my work bridges microeconomics and statistics.

Here is a graph in my paper;

Permit me to explain what is going on here. I graph a production possibility frontier (PPF) but rather than having Guns and Roses on the Y and X axis, I instead graph the student learning and leisure on the axis. Students want to learn and they want to have leisure. I argue that before AI existed that many students were NOT choosing to locate on the frontier of the PPF. They were inefficient in achieving their goals because they didn’t fully understand their optimal time allocation problem.

In the graph, the advent of AI shifts the PPF so that students are more productive. Students differ; some love leisure; vertical indifference curves and some love learning; flatter indifference curves.

The introduction of AI introduces several effects;

#1 more students move to the frontier of their PPF

#2 Students who love economics study More!

#3 Students who love leisure study less!

THE NEW POINT is that AI moves some young people from being leisure lovers to economics lovers! Preferences are Endogenous here. We as faculty have a responsibility to use this new AI opportunity to increase the % of students who have our “love of the game”!! This is an Endogenous parameter. Would Becker and Murphy frown at me here? Would they embrace the Becker and Stigler view of stable preferences? I don’t think so.

AI opens up a world of possibilities for learning if students have sufficient imagination. It is our job as educators to unlock their imaginations.